Kyte believes in an electrified future and has an ambition to operate the largest electrified fleet on the planet, making sustainable mobility more accessible for all. This year we have tripled the number of electric vehicles (EVs) in our fleet. While these vehicles were initially exclusive to our subscription customers, we will begin offering EVs for daily rentals in Los Angeles starting September 1st. 

As we celebrate this milestone, we wanted to share some reflections on electrification trends that have helped shape where we are and where we are headed. 

## A Changing Landscape
We consider four main stakeholder groups that are influencing the adoption of EVs, with big changes happening across the board.

### Governments
Government initiatives to promote the use of EVs have accelerated in the past few years. In the United States, the Biden administration [has announced](https://www.mckinsey.com/industries/automotive-and-assembly/our-insights/why-the-automotive-future-is-electric) a 50% electrification target for 2030 coupled with more strict rules on tailpipe emissions. The administration has also proposed [regulations](https://www.cnbc.com/2022/06/18/how-the-massive-ev-transition-is-starting-in-the-car-rental-industry.html) that would require EV charging stations, funded with federal resources, to be no more than 50 miles apart along interstates. Notably, Nevada, Minnesota, and Virginia have joined California and ten other states in [agreeing to establish](https://www.bcg.com/publications/2022/electric-cars-finding-next-gear) zero-emission vehicle quotas for new passenger cars by 2035.

The US has also introduced an [electric vehicle tax credit](https://www.irs.gov/credits-deductions/credits-for-new-clean-vehicles-purchased-in-2023-or-after) which is offered to consumers and businesses who purchase qualifying plug-in electric or “clean” vehicles, reducing their cost. The credits contribute to a [$400 billion](https://www.mckinsey.com/industries/public-sector/our-insights/the-inflation-reduction-act-heres-whats-in-it) in investment in clean energy - America’s largest investment to fight climate change. 

We see even stronger regulatory push in the European Union, which has drafted policies aimed at reducing the average emissions of all cars in operation by 55% by 2030 (from 2021 levels).  Going beyond this, the EU has prohibited the sale of new ICE (internal combustion engines) vehicles [starting in 2035](https://www.europarl.europa.eu/news/en/headlines/economy/20221019STO44572/eu-ban-on-sale-of-new-petrol-and-diesel-cars-from-2035-explained), setting an end date for the “ICE age” in Europe. 

### Car Manufacturers
Automakers’ investment in electrification has intensified over the past few years. Investments in hybrids and hydrogen-powered fuel cells have not disappeared entirely, but automakers are now redirecting their development efforts toward all-electric vehicles.

*Prominent players like Toyota and Volkswagen have committed a combined $250 billion by 2030 to EV and battery programs

*[Jaguar and Volvo](https://www.forbes.com/sites/michaeltaylor/2021/03/02/volvo-follows-jaguar-into-tesla-territory-to-become-all-electric-by-2030/?sh=dcbd00c69bc3), meanwhile, are setting targets to sell only BEVs (Battery Electric Vehicles) and hybrids from as early as 2025, while a [host of other manufacturers](https://www.nytimes.com/2021/11/09/climate/cars-zero-emissions-cop26.html) have set a 2040 target date

*BMW and Daimler announced a substantial $1 billion investment toward a new venture aimed at creating innovative mobility services such as ride-sharing solutions and charging infrastructure for EVs

*Volkswagen’s investments in EVs are expected to enable the company to achieve annual sales exceeding 1.4 million EVs by 2025, aiming to position Volkswagen [as the leading EV manufacturer surpassing all competitors](https://edition.cnn.com/interactive/2019/08/business/electric-cars-audi-volkswagen-tesla/)

The entire automotive sector is preparing its gears for the EV challenge. Car manufacturers worldwide are reevaluating and restructuring their traditional business models, aiming to embrace this new era where electric power replaces conventional fuels. The high costs associated with developing EVs encourage companies to seek partnerships and new innovative ways to be top in the race. 

### Consumers
Global EV adoption is expected to reach 45% by 2035 driven by falling ownership costs (supported by government incentives). The five-year total cost of ownership (TCO) for a midsize car is now the same for both BEV and ICE versions in China and many European countries.  For much of the world’s population, EVs are on track to achieve purchase-price parity with ICE vehicles by the end of the decade, a more important factor in consumers’ car-buying behavior than TCO. The continuing reduction in battery costs, which constitute around 30% to 40% of an EV’s production costs, contributes to lower ownership costs, as well as greater economies of scale due to the increased production of EVs.

One key factor for EV adoption is range anxiety - the fear or uneasiness that drivers might experience regarding the possibility of running out of battery power before they are able to reach a charging station or their destination. With the improvements in battery technology and charging networks, the impact of range anxiety has lessened, although it continues to be relevant for potential EV users especially if their travel behavior includes longer distances and areas with limited charging accessibility. With this in mind, access to private charging, switching to higher range 400-500 km EVs and the continuous improvement of public charging stations is essential for improving EV readiness.  

Public charging networks have the [potential](https://www.bcg.com/publications/2023/advancing-the-adoption-of-electric-vehicles) to enhance EV readiness through three crucial dimensions: density of chargers to reduce the distance between charging points, speed to charge to reduce the time needed to wait for a charge to be completed, and the quantity of the charging points to reduce the waiting time for an available charger. As the adoption of EVs increases, the expansion of charging infrastructure becomes essential to guarantee a charger's consistent availability, thereby mitigating any impact on EV penetration.

### Fleet Operators
This is us!

The rental market can support broader electrification goals, by providing an easy way for consumers to experience EVs before they commit to buy; many customers will have their first EV experience in a rental car. Corporate accounts are prioritizing EV rentals, due to their alignment to their own Environmental, Social, and Governance (ESG) objectives. According to Kayak’s Travel Trends 2023, searches for electric car rentals [increased 136% year-over-year](https://www.kayak.com/news/travel-trends-2023/).

Among car rental companies, Hertz is the most outspoken about its EV objectives, declaring that they own the biggest electric fleet in the world outside China. Hertz [has committed to](https://www.reuters.com/business/autos-transportation/hertz-plans-order-up-175000-gm-electric-vehicles-by-2027-2022-09-20/) purchase more than 340,00 EVs within the next 5 years from GM, Tesla, and Polestar with the overarching goal to build up one-quarter of its fleet with EVs by the end of 2024.

Meanwhile, Europcar, a French-based company, pledged to make 20% of its fleet electric or low-emission hybrid by 2024, up from the 3% owned in March 2022, which means it will need to buy [up to 70,000 cleaner vehicles](https://www.reuters.com/technology/rental-car-giants-embark-electric-after-pandemic-bonanza-2022-03-09/) in the coming two years.

## Kyte’s Plans
Expanding our EV offering has been, and will continue to be, a journey. We know our destination – the world’s largest EV operator – but it will take time and investment to get there. 

Operationally, EVs bring a new set of complexities, especially around charging. We need to ensure we have the infrastructure in place to charge vehicles between rentals and deliver a well-charged car every time. This also requires the ability to monitor charge state and battery health in the same way we monitor fuel levels and engine health on ICE cars today.

Many of our customers will be driving an EV for the first time, and we want to support and educate them to make their trip as seamless as possible. We’re developing customer resources to explain the unique characteristics of EVs, and offering charging solutions to customers while cars are on rent. 

We began this journey in early 2022, offering a handful of Teslas to our subscription customers. Launching our EV Daily Rental product in LA this September is a big milestone, but there are many more to come - expanding our EV fleet to more markets, offering a wider selection of EV vehicle classes, and ultimately becoming an EV-first operator. There will be lots to learn and improve upon along the way, and we’re excited to be taking the next step!

If you are ready to take on the wheel of an electric future, starting from September 1st, you can be among the first to experience our exciting EV fleet in LA. Whether you're cruising down the Pacific Coast Highway or exploring the Hollywood Hills, our EV fleet is designed to help you unlock the freedom to go places. If you’re ready to know more about our EV fleet, take a moment to explore our comprehensive [FAQs section](https://help.kyte.com/en/collections/5904819-electric-vehicles-ev), where we've covered everything you need to know about charging, driving, and more. 

Let's unlock the freedom to go places together!

ev-la-header.jpg

Electrifying Los Angeles: Our Launch of EV Rentals & Continued Journey

_This blog covers how Kyte’s Surfer Engineering team built the backend to develop the Surfer App, a core component of the Kyte experience that changes transportation as we know it. For context: Kyte Surfers deliver and retrieve cars from our customers. Over the last year, the Surfer Engineering team has been hard at work improving the delivery experience using the Kyte Surfer App._

## Understanding the three surfer workflows:

Once a customer has booked a car, there are numerous steps that Kyte Surfers perform in order to ensure you get the vehicle in the smoothest possible method. _Figure 1_ details the 3 main workflows they perform. The delivery flow is the sequence of actions a Surfer performs to prepare and transport a car from the lot to the customer. Conversely, the return flow involves the Surfer bringing the car from the customer back to our lot. The swap is an amalgamation of both processes, where a Surfer delivers a car from our lot and exchanges it with the customer's current vehicle.

![Surfer Workflows {caption: _Figure 1: Surfer Workflows_}{alignPhoto: center}{alignCaption: center}](https://kyte-cms-media-prod.s3.amazonaws.com/1_8d21781da8.png)

During Kyte's initial growth phase, the Surfer workflow relied on communication between our backend and the Slack API. This approach served as an excellent proof of concept, allowing us to focus on entering new markets and enabling us to scale the initial experience faster. However, as we continued to expand and add new market-specific features, we encountered several scalability challenges with the Slack approach. In light of this, we made the strategic decision to develop a dedicated Surfer app and transition away from Slack.

The first step in this transition was a complete backend service refactor. This involved building a state machine and an easily interactable service within the existing monolith. To achieve this, Kyte leveraged the [pytransitions](https://github.com/pytransitions/transitions) library and applied the State Machine design pattern to overhaul the original Surfer flow implementation. This article dives into the details of how Kyte’s Surfer Engineering team successfully achieved this significant improvement in our Surfer app development process.

## Determining the technical requirements
There were 4 major requirements that we needed to address when re-factoring the service.

### Handling multiple integrations
This refactoring process represents the first step in launching a new Surfer app that will handle Surfer-side interactions, including earnings, trip acceptance, Surfer flow, and more. To ensure a smooth transition, the migration of Surfers to our new application was carried out in multiple steps. As a result, the refactor needed to handle interactions between the original Slack implementation and the upcoming new application that we’ve built.

### Avoiding strong coupling
The Surfer flow is deeply interconnected with several Kyte services, including customer-facing notifications and our delivery/return status service. This strong coupling was not sustainable and could have led to significant challenges if there were major changes in the flow due to experimentation, market-specific feature requests, or external modifications in the services the Surfer interacts with during the flow. The best way to achieve operational reliability was by implementing a facade layer. This allowed us to make unanticipated changes, added new steps, and created an abstraction without affecting other services or teams.

### Covering all processes
As seen in Figure 1, the swap flow can be seen as a combination of both delivery and return processes, which can occur periodically when vehicle maintenance issues arise. However, the first implementation treated swaps as a completely separate entity, which resulted in the need to duplicate any changes made to the delivery or return flow for the swap flow as well. To improve our support for swaps, we were able to reuse specific steps and build the desired functionality without additional overhead.

### Building an audit trail
The final challenge to address was the absence of an audit trail. We were reliant on our Customer Support team’s help to track state changes for the operational nature of our business. Additionally, the lack of sufficient data prevented us from identifying bottlenecks across Product and Ops. Better data would provide valuable insights into improving business operations and help us continue to grow. To rectify this and place less dependency on manual human updates, we established a platform that captured the various leg transitions, which enabled us to generate event log data that contained all relevant information collected from the Surfer flow. 

## Implementing the state machine 
### Brief primer on State Design Pattern and _pytransitions_ 
Since surfers follow a series of steps during the handover process for deliveries and returns, we opted to employ a state design pattern to build a finite state machine for the backend redesign. The fundamental concept in the approach is that at any given time, the program can exist in a _finite_ number of _distinct_ states. Each unique state dictates a different behavior and actions, and upon certain conditions being met, the state will _transition_ to the next possible state until reaching the end. To implement this, we leveraged the [pytransitions](https://github.com/pytransitions/transitions) library to build state machines to implement these workflows. 

In the below code snippet, we can define the state machine model states by either passing a list of strings to the Machine instance or defining each State object and adding it in. Once added, the State object will dynamically create enter and exit callback functions which are called whenever the state machine enters or leaves that state and these functions perform all the actions of the state. The _add_transition_ function initializes all possible next state transitions and validations for a given state. The lightweight nature of the state machine library combined with its numerous extensions were beneficial during the prototyping phase. But over time, there were concerns about scalability and maintainability limitations due to the responsibilities entanglement of transitioning and performing actions for a state which would lead to poor testability and reusability. 

```
state_a_on_exit():
    #do something
    pass

# States
states = [State(name="state-a", on_exit=[state_a_on_exit]), "state-b", "state-c"]

# Initialize the state machine
machine = Machine(states=states, initial="state-a")

machine.add_transition(trigger="next", source="state-a", dest="state-b")
```

### Building an abstraction layer
To address aforementioned challenges within the _pytransitions_ library, we constructed an abstraction layer above the _pytransitions_ library and an additional facade layer over the state machine to interface with the rest of our systems. An abstraction layer is a streamlined interface to enable external services to engage with lower-level counterparts by obfuscating implementation details away. The main purpose of an abstraction layer is to promote modularity and improve code maintainability

As a company, we prioritize the capability to swiftly transition between different libraries or services based on factors like cost, performance, and resources. Dealing with a potentially outdated third-party library can pose a security risk if it hampers the upgrade of other dependencies or has issues itself. To gracefully and quickly pivot if exposed to this risk, we wrap the third-party library within our own interface, centralized in one abstraction layer. This streamlines maintenance efforts because updates and integration with the third party library are maintained within only the codebase that interfacing with the third party library and we can switch easily. 

Another major benefit of an abstraction layer is increased simplicity due to modular design. Abstraction layers are invaluable for testing purposes, allowing the replacement of specific details with alternative sets, thus isolating the test areas and facilitating the creation of test doubles. Furthermore this approach eases the reusability and code readability with our project for other engineers, as they only need to understand the abstraction layer rather than the entire system. 

For the surfer-flow case, we built our abstraction layer by taking the definition of a state from _pytransitions_, and breaking it down into 2 major components to separate the responsibilities of transitioning between states and the actions performed within a state. We created an abstract BaseStep class that handled the transitions while the BaseBehavior class is responsible for the actions performed. The flowchart in Figure 3 showcases how the whole system operates in unison. Both steps and behaviors inherit from the BaseStep and BaseBehavior abstract classes, respectively, and implement the required functions.  In terms of the state machine and _pytransitions_, as the Surfer is stepping through our flow, various actions are being performed. Examples of these actions are sending notifications to either the customer or Surfer or getting directions to the handover address and driving there. Behaviors are consistent within a certain state, but they do not care about the next or previous state.


![Kyte’s State Machine Implementation Flowchart {928x1138}{caption: _Figure 2: Kyte’s State Machine Implementation Flowchart_}{alignPhoto: center}{alignCaption: center}{](https://kyte-cms-media-prod.s3.amazonaws.com/Image_from_i_OS_2_09ba034485.jpg)


#### What is a step? 
Steps can be considered the actual states that are registered within a state machine using _pytransitions_. They are responsible for transitioning between states. Each step has a behavior registered to it based on certain parameters such as the market, the flow type, etc. _Pytransitions_ dynamically initializes the callback functions for a state, so we built the step class on top to add validations and improve testing without having to deal with the state definitions within _pytransitions_. Each step contains a transition function which lists all the various next steps the state machine can proceed to, as seen in the code snippet below. When we call the _state.next_ function, it will find the first step that meets the conditions for transitioning and then use the _pytransitions.next_ function to transition to that step. 

```
class StepA(BaseStep):
    @property
    def name(self):
        return "step-a"
    
    @property
    def transitions(self):
        return [("step-b", None)]
    
class StepB(BaseStep):
    @property
    def name(self):
        return "step-b"
    
    @property
    def transitions(self):
        return [
            ("step-c", self.condition_1) 
            ("step-d", None)
        ]
    
    def condition_1(self, **kwargs):
        # implement some condition to decide if it should go to step-c
        return False
```

#### What is a behavior?
Behaviors are the meat of our state machine design and are where all the actual work takes place. There are two potential times for a behavior to execute an action within a state: either when entering the state or when exiting it. The _on_enter_action_ and _on_exit_action_ functions allow us to create unit tests for the functions as they are now abstracted away from the generic state machine creation by the _pytransitions_ library. The pytransitions _on_enter_ and _on_exit_ callback functions sole responsibility now is to call these functions which we can test, modify, and re-use. This way, we can repurpose a Behavior implementation by registering it with unique Steps for different flows. 

Besides state actions, there is an _on_validation_ function that validates all required actions have been completed before transitioning. In the code snippet below, if the contract remains unsigned, a _DeliveryUnsignedContractException_ is raised to prevent a transition. This exception can be utilized by the front-end to notify the Surfer that they need to have the customer sign the contract to proceed to the next step.

```
class DeliveryUnsignedContractBehavior(BaseLegFlowStepBehavior):
    def on_enter_actions(self, **kwargs) -> None:
        self.facade.send_customer_notification(
            leg_uuid=self.model.leg_uuid, notification_handler_name="delivery_unsigned_contract"
        )

    def on_validation(self, **kwargs) -> None:
        rental_agreement = self.facade.get_rental_agreement(self.leg.trip.uuid)

        if not rental_agreement or not rental_agreement.is_signed:
            raise DeliveryUnsignedContractException()

    def on_exit_actions(self, **kwargs) -> None:
        self.facade.notify_signed_contract(leg_uuid=self.leg.uuid, skip_trip_channel=self.is_app)
```

#### Putting it all together
Referencing _Figure 2_ once more, each step is registered with an individual behavior in the _BehaviorFactory_. Subsequently, the steps are registered to corresponding state machine flows (i.e. delivery flow, return flow, and swap flow) within the state machine factory. Upon receiving input of the current state and based on the transition parameters, the state machine transitions to the next step. Our implementation has now split the responsibilities of a state to simplify testing each unit of the system. Now we can effortlessly make market-specific changes and fulfill new feature requests within a behavior without needing to touch the state transition logic within steps by taking advantage of the abstraction layer we built. 

## Conclusion
The endeavor to improve our surfer flow was a collaborative effort involving Engineering, Product, Communication Support, and Leadership. Rewriting an entire service is a delicate balance between prioritizing the values and components most crucial to the company and those that matter most to the customers. The _pytransitions_ library laid a significant amount of groundwork to enable us to completely refactor the surfer flow service. Building upon this foundation, we followed the principles behind abstraction layers and state design patterns to build and encapsulate a robust service within the monolith. Although our approach enabled us to work efficiently on the app and provide a reliable and positive experience for our Surfers, there is still much work to be done to revolutionize the Surfer experience and transform transportation as we know it. As we expand into new markets and continue to grow, we anticipate encountering new challenges, identifying further improvements, and undertaking larger projects.

viktor-kiryanov-zQjQaEiAafk-unsplash.jpg

Executing State Machine Transitions with PyTransitions

When was the last time your car rental company shared its technology roadmap? To elevate the experience of delivering rental cars, Kyte must openly share our culture of continuous improvement and transparency, building both technology and operations with you the customer in mind.

## Background
What is Kyte? Kyte delivers rental cars directly to you for any trip longer than a rideshare. We currently serve our customers across 13 markets in the United States. Since Kyte was founded in 2019, we have differentiated ourselves in quality and convenience in a number of ways. We don't have rental counters or stores and make technology and operational investments to drastically elevate the rental experience and keep costs low for our customers.

In 2022, we launched many product features: 

- Self-service features allowing customers to extend their rental or add an additional driver 
- Our surfer app to improve customer service and the surfer experience 
- The Kyte subscription product
- Tipping for surfers
- Live location tracking of the rental delivery

![live-location-view](https://kyte-cms-media-prod.s3.amazonaws.com/export_blog_live_location_ea56571059.webp)

In addition, we remedied regularly occurring internal tooling outages to improve our employees’ ability to assist customers

We’ve been fortunate to land ABS financing from [Goldman and Ares](https://www.prnewswire.com/news-releases/kyte-secures-200-million-asset-backed-financing-facility-from-goldman-sachs-and-ares-management-to-grow-fleet-amid-unprecedented-transformation-of-the-transportation-industry-301509261.html), furthering our ambitions to become the world’s largest fleet operator and were successful in completing a [Series B fundraise](https://www.phocuswire.com/on-demand-car-rental-startup-kyte-raises-60-million-for-expansion) during a very difficult time in the market. Overall, it has been an amazing few years at Kyte!

While these product and operational milestones definitely show progress, one of our values at Kyte is to get better every day, and our technology is no exception. Over the next 18 months, Kyte will invest in a technology roadmap that further improves self-sufficiency for our customers, increases trust and transparency on our platform, creates frictionless experiences, and allows for our team to proactively solve customer problems.

## Self-Sufficiency for Customers
Self-explanatory product flows allow us to support our users without increasing our support volume and personnel - another way Kyte enables convenience and cost efficiency for customers. An increasing number of Kyte customers are finding convenience in one of our newly-released features: Trip Extensions. Using the [Kyte app](https://kyte.com/download), you can extend your trip without having to contact support. Just simply modify the date, time, or location with ease. Throughout the year, we will also introduce an increasing amount of flexibility around car handovers by expanding our ability to deliver outside of our service areas as well as other options.

For example, we’re piloting pickup-at-lot, which will allow customers to select their preference of self-pickup or delivery. We expect that giving this choice to customers will lead to a more convenient experience while giving you, the customer, a higher degree of flexibility around forming your Kyte experience to your needs.

![edit-trip-example-blog.png](https://kyte-cms-media-prod.s3.amazonaws.com/edit_trip_example_blog_5000669a67.png)

## Trust and Transparency
For most rental car experiences, common complaints are due to poor car cleanliness or lack of transparency around fees for things like tolls and citations. Kyte cleans and maintains its cars through a workflow that is executed at the end of every trip. Our customers enjoy low mileage, clean vehicles at a reasonable price. With the launch of our surfer app, we plan to go deeper on the quality of our cleaning and maintenance workflows to ensure the highest quality experience.

This year, Kyte plans to simplify the traditional car rental receipt structure, by providing full transparency about what fees customers are charged for a given service. Whether you zipped through a toll station with ease or received a surprise parking ticket during your trip, Kyte will provide the ultimate transparency. We will also simplify the receipts you receive to allow for easy business expenses and to maintain a simple to understand booking experience that is best in class.

Rental agreements are one of the most complicated topics of the car rental business – we will further increase trust by creating documents that are crisp, upfront, and easy to understand. We plan to build an easy-to-understand and up to date documentation for your rental including any coverage or limitations that are included. Customers will have full access to their past rental agreements in case of need for reference or in the situation where a damage or claim arises.

## Frictionless Customer Experience
Part of Kyte’s value proposition is to decrease the overall friction involved in getting a car. Our technology will support the majority of edge cases encountered in the [world of atoms](https://kyte.com/blog/bits-and-atoms), and assist our market leaders with marketplace efficiency in dispatching our surfers, managing our lots, and empowering customers to solve their own problems. 

To provide a 5-star delivery, our matching algorithm will advance to provide global optimizations so that auto-matching becomes the norm for surfers, ensuring fewer trips result in bad experiences and that the ideal surfer is matched or re-matched to an available trip with ease. We will focus on adding re-matching and using global optimization to build surfer schedules so users and surfers experience the highest quality scheduling automation available to date.

## Proactive Problem Solving
Kyte’s internal operations have a direct impact on your customer experience. Unresolved problems can lead to issues at our lots that sideline vehicles and create a bottleneck for us to provide you with your transportation as effortlessly as possible. By continuing to invest in our Dispatch and Support tooling we can flag small problems (e.g. delays, scheduling issues, etc.) before they turn into larger, compounding issues in our markets. 

![surfer-workflows](https://kyte-cms-media-prod.s3.amazonaws.com/export_surfer_workflow_blog_7ac3251600.webp)

Kyte can also streamline the traditionally nightmarish claims process by using automated condition reports and one-click claim creations to cut the processing time. Kyte can catch and resolve any vehicle issue and provide the fastest response to our customers with these tools. Investing in this internal tooling allows us to provide the highest quality experience at the lowest cost for you.

## Conclusion
We’re excited to share more over the coming months about more of our Technology plans. These are just a few of the exciting updates we are focused on, we have SO much more to share with you!

mark-konig-ECGv8s2IPG0-unsplash.jpg

When Was the Last Time Your Car Rental Company Shared its Technology Roadmap?

Kyte is a new generation transportation company. We bring you a car just when you need it, for as long as you want, thus unlocking your freedom to go places without having to wait in lines or waste your time waiting at the rental counter.

At Kyte, we perform numerous operations offline. Once a user books a car via the Kyte website or mobile app, a number of tasks need to be completed before, during, and after the rental starts. An example is multiple reminders and notifications sent at predefined times or at a predefined cadence. Historically, multiple ECS cron jobs have been added for the purpose of doing some asynchronous processing in a batch-like fashion. Kyte Engineering has found that over time this approach presented various problems related to tracking, scaling, and maintaining the flows. As our load volume kept increasing, the problems above were becoming more and more pressing, so we decided to start investigating workflow as an architectural solution.

## Problems

The sequential batch processing approach does not scale. Each booking’s processing is delayed by the processing of all bookings ahead of it. If the task is time sensitive, at the time it’s executed it may no longer be relevant or may delay some other critical tasks it triggers. Additionally, the ability to retry a failed task is limited, because each retry will further delay processing of other tasks. The logic to process tasks for multiple bookings would be:

- give me all bookings that satisfy certain criteria;
- loop through the bookings and perform the specified task on each.

![Diagram 1 {caption: Before workflows.}](https://kyte-cms-media-prod.s3.amazonaws.com/blog_temporal_before_workflows_90856edaf9.jpeg)

Parallelizing the processing may solve the scalability issue above, but comes with other issues:

- design and maintain parallelization;
- keep track of retries;
- persist data and state across restarts;
- recover from individual failures.

## Solution

A widely accepted architectural industry standard intended to solve the stated problems is a workflow engine. A workflow engine takes care of retries, statefulness, and scalability while allowing the developer to focus on the business logic. Additionally, in a workflow paradigm we can implement a workflow of a certain type and launch one instance of it per entity (in our domain, this can be a rental booking, a trip leg, a subscription, etc.).

![Diagram 2 {caption: After workflows.}](https://kyte-cms-media-prod.s3.amazonaws.com/blog_temporal_after_workflows_d0e95ffcb3.jpeg)

The setup above allows us to

- configure a retry policy executed on each entity (e.g. booking, user, trip leg) where the retries are fully handled by the workflow engine;
- design a process as a set of atomic actions (activities) that always only run once on success;
- not worry about the state of the flow across restarts;
- easily track processing of a specific entity via the workflow UI:

![Diagram 3 {1000x140}{caption: Workflows chart.}](https://kyte-cms-media-prod.s3.amazonaws.com/blog_temporal_workflows_chart_4807a30d47.png)

## Why Temporal?

We evaluated a number of existing workflow engines based on a set of weighted criteria, such as availability and maturity of Python SDK (as we are currently mostly a Python shop), documentation availability and completeness and server setup and maintenance cost. Temporal won as an open source yet stable and extensively used product with a highly involved and expanding community, great level of support, and a Python SDK in development nearing GA.

Unlike some other workflows out there (e.g. Airflow), Temporal is not designed as a data processing engine. A Temporal workflow is not a pre-defined pipeline or a DAG. A Temporal workflow is a function written in a programming language of choice and can be event driven (i.e. can listen to signals), which is extremely important for Kyte’s business case where a different course of action or a different code path need to be taken depending on the real-time, real-world situation.

## Kyte Setup

We started with a self-hosted Temporal cluster as a proof of concept. However, having limited infrastructure development resources, we decided that maintaining our own EKS cluster (including deployment, upgrading, scaling, user authorization) was not a task we could afford spending enough time on and we’d become more efficient by delegating that work to Temporal, so we soon moved to a managed Temporal, hosted by [temporal.io](https://temporal.io).

We (Kyte) deploy and manage our Temporal workers (client processes running the workflows) while Temporal fully manages and maintains the server user authentication and authorization, server versioning, scaling and tuning, and environments.

![Diagram 4 {caption: Kyte's Temporal setup.}](https://kyte-cms-media-prod.s3.amazonaws.com/blog_temporal_kyte_temporal_setup_d06ec79d11.jpeg)

## Monitoring

The Temporal SDK emits a number of useful metrics via Prometheus by default. We initialize the metrics on startup as follows:

```
prometheus_http_port: str = os.getenv("TEMPORAL_PROMETHEUS_HTTP_PORT") or "18081"
_init_metrics("0.0.0.0:" + prometheus_http_port)
```

We collect the metrics in DataDog by scraping Prometheus endpoints. The metrics are used to create dashboards as well as set up monitors and alerts. We currently have the following types of monitors:

- workflow heartbeat monitor (based on the _temporal_workflow_completed_ metric);
- workflow activity latency is too high (based on the _temporal_activity_schedule_to_start_latency_ metric);
- workflow failed (based on the _temporal_workflow_failed metric_).

## Local Testing

When making changes it is useful to quickly test the orchestration part of the workflow without the overhead of connecting to Temporal. For local testing we are using _Temporalite_, a distribution of Temporal running in a single process with no run-time dependencies.

Our _docker-compose-temporal.yml_ contains the following:

```
version: "3.5"

services:
 temporalite:
   container_name: temporal
   build: https://github.com/temporalio/temporalite.git#main
   networks:
     - kyte
   ports:
     - 8233:8233
     - 7233:7233

networks:
   kyte:
       name: kyte_default
       driver: bridge
```

## Pitfalls

### Worker Scalability

As we started scaling out, we noticed that our Workers could not handle the load. The workflow in question was designed to estimate and flag the risk of delivering or picking up the car on time, so a workflow per a rental trip leg was spawned (most rental trips have at least two legs, delivery and return). That summed up to about a thousand workflows per day each running a few activities in parallel. Despite not a very high load, some workflows seemed “stuck” while waiting to get executed. Adding resources to Worker processes did not seem to solve the problem.

A discussion with Temporal engineers revealed a configuration issue. A Worker object created via the Temporal API is configured to handle a certain number of parallel activities (in the beta version of the Python SDK the default is low). With a large number of workflows each starting a large number of activities simultaneously, Workers reach the limit and stop asking the server for work. Increasing the max_concurrent_activities setting when creating Workers solved the issue without adding computing resources.

### Long-running Activities

Because a car could be rented for days, weeks, or months, some activities mapped to bookings will also be long-running as they run for the period of the booking. An example is an activity continuously checking for the status and location of the car.

If a worker goes away in the middle of executing an activity, there are two ways the server will detect that and hand the activity to another worker:

- when the start_to_close_timeout is exceeded (i.e. when the activity takes longer than supposed to);
- when the heartbeat_timeout is exceeded (i.e. when the Worker stops checking in with the server using the heartbeat API).

Because we do a full deployment on every code change, our Workers are frequently brought down. In the scenario where a Worker started executing an activity and a subsequent deployment replaced the Worker, without a heartbeat, the Temporal server would wait for the _start_to_close_timeout_ (could be days) to fail and re-run the activity. Adding heartbeats to our activities in combination with heartbeat timeouts resolved another issue of seemingly “stuck” workflows.

### Non-determinism and Backward Compatibility

A workflow’s code must be fully deterministic. E.g. if a Worker dies while executing a workflow and a new Worker picks up the workflow, the workflow must be replayed to the point where it previously stopped. In other words, it must do the same exact thing on every execution. If the workflow code is changed between deployments, the changes may lead to non-determinism errors. There could also be code in the workflow itself introducing non-determinism (e.g. calling _datetime()_).

The right way to deal with this is to use the Temporal versioning API for any non-backward-compatible changes and to always check the workflow code for any non-determinism issues before deploying. It is possible to test for non-determinism using the Temporal workflow Replayer API.

However, if you already have hundreds or thousands active workflows in deployment failing with non-determinism errors, what do you do? One way to mitigate that is to reset all the workflows in a batch fashion. Our workflows are designed to be fully idempotent. An example is the deposit charge and refund workflow. If a workflow charges the deposit for a booking and is restarted, it will detect that the deposit has already been charged and will not charge again.

We use a _tctl_ (Temporal server command line tool) script to start workflows anew.

```
workflows=$(tctl --tls_cert_path $temporal_api_cert_path --tls_key_path $temporal_api_key_path --address $temporal_server:$temporal_port --namespace $temporal_namespace wf listall --workflow_type $workflow --open)

workflow_list=$(echo ${workflows} | awk -F " *" 'NR>1 { print $4}')

echo $workflow_list > workflowlist

tctl --tls_cert_path temporalsdk.pem --tls_key_path temporalsdk.key --address $temporal_server:$temporal_port --namespace $temporal_namespace workflow reset-batch --input_file workflowlist --reset_type FirstWorkflowTask --reason "Ensure backward compatibility"
```

The good news is that the Temporal engineering and support teams are extremely helpful and responsive and as we keep rolling out Temporal usage to new use cases, we gain more Temporal knowledge so the experience is smoother with each new workflow.

blog-temporal-main.jpg

Kyte's Journey of Temporal Workflows

## Operations and Assets in Silicon Valley

Whoever decides to start fundraising for a company that involves (a lot of) physical assets and operations soon realizes that there is an unwritten law, amongst Silicon Valley investors, that you should avoid (in-the-real-world) operations at all costs. Operations-heaviness is considered dangerous waters that you should avoid. And given that operations-heaviness is considered bad, then asset heaviness is considered even worse. “Being asset-heavy” is considered taboo and hazardous for many traditional Silicon Valley tech VCs. You might as well write “don’t invest” on the first slide of your pitch deck.

At Kyte, we have a contrarian and provocative view on many of these elements. So, in the following, we’d like to unpack a few core elements around tech startups, operations, and assets in Silicon Valley.

### Rewarding but hard problems

When we started Kyte, one of the core driving forces was that we were incredibly excited by solving real problems in the real physical world. We believe solving problems in the physical world is the most exciting and interesting space to be in - you impact people's lives in a very direct and tangible way, you make the real world more efficient, and you help the world use its limited resources more efficiently, and thus you contribute to a sustainable future with more optimal resource allocation.

On the flip side, these are also the hardest problems to solve. There are so many companies in the “Uber for X” marketplace graveyard that infamously failed (valet services, delivery services, concierge/private services, etc, etc). Let’s unpack what exactly is so hard about them and unpack some “myths” that can explain this aversion of Silicon Valley investors. Then, let’s define how to crack these beasts.

### Unpacking what is really hard and how to do it right

- Dealing with operations and assets is hard, really hard, because things in the real world break, they depreciate, they become dirty, data is noisy, and things are by default never “clean” (it's about cars, assets, but also about sensor data, signals, measurements, etc). Everything is 10x messier than in the purely digital / software world.
- Dealing with operations and assets is traditionally not a core competency of Silicon Valley. It’s not what a lot of Tech VCs understand. And what you don’t understand, you don’t invest in.
- Gross margins are oftentimes slim, CAC can make or break your company, … and ultimately it’s expensive to add humans into an equation. (In particular, in developed countries with a high salary level.) Physical goods have different profit equations than pure software products. A lot of people underestimate these mechanics and get deceived by amazing initial consumer reviews … of course I would buy an on-demand service that delivers cute dogs to my door for petting if it’s cheap and convenient, but “product market fit” is not really product market fit if the economics don’t work. Afterall, anyone would buy a dollar for 95 cents.
- Asset heaviness often means a different form of capital - debt capital, loans, … these are different types of capital than venture capital. Brutal covenants, very different terms, etc. Banks, with very different risk profiles than venture capitalists, protect themselves in very different ways than venture capitalists. Banks can never lose, VCs can. It’s scary at first;it's a different type of being on the hook. It's a different ballgame than an easy multi-million dollar SAFE round and then you go back into your garage to write code.

**The solution**: If you are building a company with ops and asset components, you have to do it right and you have to operate at a different (much higher) level of excellence than other companies.

Here are some core “recipes” if you still decide to tackle such a beast:

1. Gross margins are lower, so focusing on unit economics is more important than ever; you have to build a really healthy / sound business from the beginning. You cannot let amazing consumer reviews in the early days deceive you into a world of unsustainable economics in the long run. And you cannot explain away bad economics with “economies of scale” in every single aspect of your business. Yes, economies of scale are real, but they are almost always pitched as more shiny and more important than in reality.
2. You have to earn your growth. A good framework for earning your growth is to focus on driving a single market to profitability. Focus on n=1, then n=2, and then scale to n=10. Get your local economics into good, better, great shape, and then scale from there.
3. Build in sequences and don’t underestimate complexity at any point in time. You have to earn taking on more of the ops value chain. Kyte started on the shoulders of rental car companies, and now, 3 years later, we literally own and control hundreds of vehicles. But we are still not doing EVERYTHING ourselves. We are not repairing cars ourselves. We do not own auto shops …. . We are very cautious with the pieces of the stack that we take on.
4. Most importantly, you have to hire great people that are really good at doing all of this. Think from first principles and don’t think that the SF Bay Area has all of them. You both need people that have seen crazy stuff… accidents, lawsuits, real stuff going wrong, as well as the best raw talent that fits your company. And realize that you have to be so good, so freaking good, to build this business. You have to be scarily good; it can’t be overemphasized enough. We are looking for incredibly good people. People with raw intellectual horsepower, humbleness, the attitude to reflect and learn, the drive to tackle the hardest and scariest problems with grit, persistence, and pure force, and ambition to win and go big.
5. And lastly, don’t think it's easy. At Uber, people always said: Nobody just lets you build a billion dollar company. Nobody is handing it just to you. You have to fight for it. It’s true, it's incredibly hard, so you have to hire for this level of resilience, persistence; you have to hire people who, when getting slapped in their faces, don't lose their persistence and continue fighting back with all their energy, or even more.

## Creating true trillion dollar value

### Marrying bits and atoms

When we started Kyte and even more when we started to scale it, we realized we need to make a statement about operations vs tech. We cannot implicitly take the answer to this question for granted and assume it works. We have to make it explicit.

Here, in a nutshell, is the type of company Kyte is:

**We are 100% a tech company. We are 100% an ops company. We are both.**

Note, it’s not an “either … or”, it’s an “and”. This is super important. This means that operators are equally valued and respected at Kyte as engineers, product, growth, data folks, etc. Operational excellence and engineering/product/ growth/data excellence are both target states that we are striving towards. It’s the intersection of software and operations, data and real assets in the real world, across bits and atoms. We’re building the real on the ground operations, the turn-arounds, the maintenance, the customer operations, the asset management. And at the same time, we need the digital world where it's about the bit. Creating magical products for users and Kyte surfers, efficient growth engines, seamless in app experiences, building algorithms that literally dispatch hundreds of trips per day for Kyte. These bits manage hundreds of vehicles across Kyte’s markets. It’s complex and you can only manage it with the best software stack possible.

## Why this is valuable

There are very few players in the space that can own the full spectrum, who can be excellent in BOTH. Google can't do this (and they have shown that they cannot do operations and assets really well, e.g. that’s why they are using Avis to service their fleet of Waymo cars).

Hertz can’t do this (they’re an ops-only player and have shown they are not able to build a great technology enabled product experience). It's funny that Uber uses the phrase “Bits and Atoms” and claims to be a company that understands assets. I would say, they don’t. Examples here involve the fact that the bike and scooter programs didn’t take off for Uber, as well the infamous “Xchange leasing” program (the program where vehicles were rented to Uber drivers) didn’t make it. Uber loves to be a technology marketplace, and Dara literally came from Expedia so the “Marketplace”-DNA makes sense. Maybe, with some of the old leadership, Uber would have the potential to go after an Amazon sized opportunity and seriously take on the asset and operations game, but it doesn’t look like they are willing to go after this anymore.

What we at Kyte are doing is super hard. But if we get it right, it’s incredibly valuable. That’s also why Amazon is a $1.7 trillion company. They marry bits and atoms in a beautiful and complex way - both for core Amazon as well as for AWS: Amazon is literally operating hundreds of fulfillment centers with ten thousands of vans, as well as their own data centers across the globe - with real assets on the ground, and with the best software stack orchestrating this engine. Building and scaling out this engine and infrastructure is extremely hard at first and requires an insane level of operational and technological excellence, but once you build it, it’s one of the strongest infrastructure moats of our time. There are very few companies that get this right. However, if you tackle bits and atoms in the right way, you will make it.

## Conclusion

At Kyte, we are bold to take on one of the hardest problems out there. We don’t run away from operations or assets. Instead, we embrace dealing with the real world because it is hard, not a lot of companies are good at it, and because we know that if we get it right, it is tremendously valuable. We are diving head first into it.

We are looking for the best operators and the best engineers, growth, product, data, supply, finance, BD folks. You will be able to solve hyper complex and massive problems on a truly global scale. We are building some of the most complex logistics algorithms, growth engines, fleet intelligence systems, etc. that you can imagine. **Everything involves bits and atoms at Kyte - and orchestrating this is our secret sauce.**

blog-bits-and-atoms-roads.jpeg

Bits and Atoms: Building a Trillion-dollar Tech Ops Company

Need a car? Don’t want to stand in a long line at the airport? Kyte comes right to you.

Seattle is our eighth market and we could not be more excited to give Seattlelites a flexible and convenient option for exploring the city and natural surroundings. From downtown Seattle to nearby neighborhoods such as Ballard, Fremont, Capitol Hill, SLU and Columbia City, our professional fleet of fully serviced and sanitized rental cars are available for delivery and pick up at user’s doorsteps.

![Seattle Map. {1530x1324}{caption: The areas of Seattle where you can get a car delivered right to you.}](https://kyte-cms-media-prod.s3.amazonaws.com/blog_kyte_expands_seattle_map_117599a0a3.jpg)

Seattle residents and travelers can now easily head to Columbia Gorge for the weekend or spend the day exploring all of the city's historic tourist spots by booking their car rental through the [Kyte website](https://kyte.com) or [mobile app](https://apps.apple.com/us/app/drive-kyte/id1456424048).

The way it works is simple — just select your preferred vehicle class and request a time and location for delivery and pick up that is most convenient for you. A “Kyte Surfer” takes care of the rest, effectively eliminating the long lines and paperwork.

We’re thrilled to give Seattle locals an alternative to currently existing transportation options. Our own Seattle resident and Chief Operating Officer, Chandra Morando, spoke on the significance of our recent launch:

> "One of the most pivotal points in my life was moving to Seattle in April of 2016. I was completely overwhelmed by the beauty of the Pacific Northwest; the snow-capped mountains, the seemingly endless waterways, the prominent presence of trees throughout even the densest parts of the city. I have had the privilege of living in some amazing cities, but this was unlike anything I’d experienced before.
>
> I moved to Seattle without a personal vehicle. I took the light rail from my apartment in Columbia City to my office downtown. I relied on public transportation as my main way to move around and explore the city, and it’s still my preferred method of transportation. As great as Seattle transit is, I found myself limited when it came to exploring the amazing enclaves outside of the city limits. My options were high-priced car shares or ride-shares or inconvenient traditional car rental facilities. I often found myself opting out of experiencing everything the PNW has to offer, rather than go through the hassle of figuring out transportation. This is a tradeoff Seattlites no longer have to make.
>
> I could not be more excited to launch Kyte in Seattle and help my neighbors more freely experience their surroundings. At Kyte, we deliver new, clean vehicles right to your door. Now, it’s more convenient than ever to hike Mt. Rainier, experience the majesty of Snoqualmie Falls, or explore the breathtaking San Juan Islands."

Since launching Kyte in 2019, our mission of replacing car ownership with access to vehicles on-demand has increased along with the need for competitively priced, reliable transportation. We’re making it surprisingly easy to get a car — and we’re excited to continue expanding our services to more cities throughout the U.S. in 2021 and 2022.

blog-kyte-expands-seattle-skyline.jpeg

Kyte Expands to Seattle

Today I’m excited to share that [Chandra Morando](https://www.linkedin.com/in/chandra-morando-1767b256) is joining Kyte as our first Chief Operating Officer, where she will oversee all operations, partnerships, expansion, and the rapidly growing list of markets.

Kyte is at a turning point of maturing from a niche product, to emerging as a global leader in the Car-as-a-Service market. With over a decade in the automotive and transportation industry, no individual is better equipped to help lead Kyte into this new era than Chandra’s unique combination of experience ranging from the biggest fleet management company on the planet to the notoriously [fastest growing startup of all time](https://qz.com/1305719/electric-scooter-company-bird-is-the-fastest-startup-ever-to-become-a-unicorn). Rarely does an executive cover the breadth of operating experience in the real world and the strategic acumen to grow complex operations into a multi billion dollar P&L. Chandra impresses all who have the pleasure of working with and for her; she’s built an impressive following, and has proven herself as an authentic, kind, principled and relentless leader in the ever-changing and challenging transportation industry.

Before joining Kyte, Chandra led the North America Operations for Bird Technology, a pioneer in personal mobility and the notoriously fastest growing technology startup at the time with an asset-heavy, operationally-complex model and a ridiculously ambitious mission. As part of the predominantly Ex-Uber executive team, she led her team of well over 200 operators into an aggressive North American expansion and through an uncertain peak-pandemic environment with maximum speed and execution focus - lessons that Chandra will carry to further accelerate Kyte’s expansion plans.

Prior to Bird, Chandra led bikeshare giant OFO’s expansion into the Western Region, fueled by their Alibaba-led $866m Series E.

Prior to that, Chandra helped establish Zipcar for half a decade as the leading car sharing player at the time - where she led the Pacific NorthWest as Regional General Manager. Zipcar, founded in 2000, was ahead of their time as they provided a growing population of urban dwellers with a transportation option that, at the time, was a better way to access cars and helped make cities better places to live, further fueling Chandra’s passion for solving urban transportation.

Chandra gained early operating experience in the transportation industry at Enterprise Holdings - the biggest fleet management company in the world. With a fleet of well over 1,700,000 cars and over $22 billion revenue in fiscal year 2020, Enterprise showed that a well diversified car business in local markets can be profitable and thrive, despite their arms-length approach to technology and innovation.

> “I’ve dedicated most of my career to providing people with better ways of moving around their cities while reducing the burden of personal car ownership, and removing vehicles from our overly congested urban centers. I could not be more excited to continue this mission with the talented team at Kyte, as we give people the easiest way to get a car when they need one. The Kyte experience is unlike any other, and I cannot wait to bring it to more people in more cities.

Chandra is joining us during a critical time in the company growth trajectory where we’ve reached an inflection point. In the past 12 months, Kyte grew well over 1,000% as more people are prioritizing car services over car ownership, despite a largely depressed transportation and Car-as-a-Service market that separated the wheat from the chaff.

As our North American markets are emerging from the pandemic, the way is cleared for Kyte to expand aggressively and, for the first time since inception, capitalize on the tailwinds caused by pent-up demand and an auto market that’s been dramatically impacted by global supply chain disruptions. We further strengthen our unit economics, and accelerate our growth into more markets across North America, while Chandra will help us mature as a company, further solidify our operations, and absorb the complexity of rapid expansion as we capitalize on the massive opportunity ahead of us.

As a strong and experienced executive and the second woman to join Kyte’s leadership team, Chandra will help further drive our agenda of diversity and inclusion and help attract diverse ‘A+’ talent to join Kyte’s mission.

By making it surprisingly easy to get a car, Kyte is giving people better access to fewer cars. With our relentless commitment to customer centricity, technology-powered operational excellence, and capital efficiency, we’re transforming the way people access transportation in growing, dense urban environments. After becoming the fastest growing company in the Car-as-a-Service space and positively impacting tens of thousands of urban dwellers across North America, we’re still recognizing we are at day 1 of the journey ahead of us.

Kyte is growing - if you’re interested in giving more people access to fewer cars, we’re hiring across every department. See [here](https://jobs.lever.co/kyte) for more info.

blog-welcoming-chandra-morando.jpeg

Welcoming Bird’s Ex-VP Operations Chandra Morando

Alongside lifting restrictions and widespread vaccination efforts, New Yorkers and travelers to the Tri-State area will finally have the freedom to explore the Big Apple and all of its surroundings without having to go miles out of the way to pick up and drop off a rental car or taking a taxi to the car rental location. Starting May 1, 2021, Kyte’s professional fleet of fully sanitized and serviced vehicles will be available for delivery and pick up in Mid to Lower Manhattan. With Kyte, customers can easily book their preferred vehicle through the app or website and a Kyte Surfer will drop off and pick up the car at a convenient time and location as requested.

Kyte will make it surprisingly easy for all New Yorkers to get a car anywhere in Manhattan with no lines and no paperwork.

![Manhattan map. {846x856}{caption: The areas of Manhattan that we are delivering to in our initial launch.}](https://kyte-cms-media-prod.s3.amazonaws.com/blog_kyte_expands_nyc_map_2c010bde5e.jpg)

“We wanted to offer a better car rental experience in New York City without having to go all the way to JFK, La Guardia, or anywhere for that matter,” said co-founder Ludwig Schoenack. “With Kyte launching in New York City, our customers won’t have to rely on taxis or public transportation to get around and will instead have the freedom and flexibility to go wherever and whenever they please'', he added.

With flexible options for long-term car subscriptions and short-term rentals, a weekend trip to the Hamptons or a day out seeing the city’s most iconic tourist spots like Times Square and the Brooklyn Bridge are all a few clicks away.

After Kyte’s success during COVID-19 times in Boston, Washington DC, San Francisco, and Los Angeles, the San Francisco-based company decided to open up in the New York market with more cities planned for 2021 such as Miami, Seattle, Chicago, etc. “A year ago, we beta-launched in our second city, Los Angeles, in the beginning of the pandemic. Since then, the need for safe and reliable transportation has only grown and shows no signs of leaving,” says Schoenack.

## About us

Kyte is a Series B technology platform that brings magic and seamless delivery into the experience of getting a car in cities for any trip longer than a rideshare. We unlock the freedom to go places by committing to relentless customer centricity, technology-powered operational excellence and capital efficiency. The sky's the limit when it comes to Kyte.  

At Kyte, we love what we do and we believe our diverse team of ridiculously ambitious individuals is what makes our company unstoppable.  We are a transparency first, egoless, and down-to-earth group of people, scattered across the globe, with a true passion for the Kyte mission. No matter where you are or what you are working on, you will have a significant level of ownership and impact on our company's trajectory and growth into the future.

Help us unlock the freedom to go places!

blog-kyte-expands-nyc-skyline.jpeg

Kyte Expands to NYC

Today we couldn’t be more excited to share that Kyte has raised $9M from DN Capital and Amplo VC, with participation from former Uber executives including Ed Baker (former VP of Product) and Jörg Heilig (former VP of Engineering), Josh Mohrer and William Barnes as well as innovators in travel and mobility like Toby Sun (co-founder, Lime) and Terry Jones (co-founder, Kayak and Travelocity). The funding will allow us to accelerate our mission to give people better access to fewer cars across new geographies. We’re already in San Francisco, Boston and Los Angeles.

We built Kyte to address a major gap in the mobility market. Uber and Lyft are great solutions for a single ride delivered right to your doorstep, but no such solution exists for when you need a car for a few days or weeks. People are forced to trek to a car rental branch, which is usually located by the airport or in a location outside of the city-center and then wait in a long line in an enclosed space for their rental. Some have turned to carshare platforms, where one is left to locate an uninspected (and often uncleaned) car on their own and fend for themselves, making quality and reliability highly unpredictable.

Enter: Kyte. We bring the same amazing experience to renting a car that people have come to expect in every other part of their life. Whether it’s ordering a package on Amazon, a ride on Uber, or a meal on Uber Eats, doorstep delivery has become table stakes. Kyte reliably delivers a quality car to the customer with just a few taps in our app. What makes the magic possible all comes down to our fleet operations platform matching people with cars in the right place at the right time, ultimately orchestrating vehicle movements across dense urban areas. We partner with existing professionally managed fleets, allowing us to focus solely on perfecting the consumer experience and building technology - not on managing assets.

“We believe that Kyte is going to be a core building block to the future of mobility,” said Steve Schlenker, Co-Founder and Managing Director, DN Capital. “The pandemic has accelerated the transformation of cities and consumer behavior with respect to transportation. Kyte’s unique operations layer facilitates this transformation while providing a level of service and convenience that other solutions fail to meet. We are excited to help the Kyte team, with our investment, to accelerate delivery of this solution to more people in more places.”

One of the things we’re most proud of is that when customers try us, they love us. Despite fast growth in the past months, over half of our bookings come from recurring users and our Net Promoter Score of 85 is unseen in the industry. We’ve also watched as people have become increasingly reliant on our product during the pandemic and have grown 400% since March. This has been fueled by a decrease in plane travel and a resurgence in the road trip and weekend getaway. Kyte offers one of the cleanest and safest ways to access a car for these adventures.

From customer feedback we understand that Kyte - in combination with ridesharing and micro-mobility - is truly building towards a better and more efficient transportation system that focuses around people and not cars. A world where people don’t have to own cars and can have better access to fewer cars because cars are stored in central hubs and are moved to where they are needed through Kyte. Owning a car is incredibly expensive and cumbersome, plus it results in vehicles taking up precious curb space that could be better used for things like parklettes and outdoor dining. In order to build towards this human-centric vision, Kyte unlocks the most compelling and easy way to get a car for consumers. To achieve this, we need to grow and be universally available in cities spanning the country. We can’t wait to step up to the challenge and make this a reality.

We started Kyte after working at Uber, BMW, and McKinsey. We grew the team with people from category defining companies like Expedia, Bird, Apple and Facebook - and we continue to grow. If you’d like to join us on our mission, check out our [job openings](https://jobs.lever.co/kyte). If you’re interested in trying out Kyte, book with us [here](https://kyte.com) or [download our app](https://apps.apple.com/us/app/drive-kyte/id1456424048) today in the Apple Store. If you’re a member of the press that would like to chat with us about today’s news or the industry more broadly, email hello@drivekyte.com.

blog-kyte-founders.jpeg

Kyte Raises $9M to Expand Doorstep Car Delivery in the US

We are super excited to officially announce the opening of our new (currently virtual) Engineering & Product Office in the beautiful capital of Bavaria. Here, we’d like to take a few moments to reflect and shine some light on the reasoning and what that means for the entire Kyte team.

## Why we chose Munich

First of all, there are a few very obvious reasons:

- We are a true technology company and there is simply amazing technology talent in Munich given the presence of two elite universities (Technische Universität and Ludwigs-Maximilians-Universität) with world-class computer science and engineering programs at both of them. In addition, Munich has become a true European tech hub and more and more tech companies have discovered Munich as their European office location.
- There is great and diverse entrepreneurial talent in Munich, in particular in the high tech sector (greetings to our dear friends at Luminovo, Celonis, Fernride, Personio, Demodesk, ParkDepot, …). Incubators and university-affiliated programs such as Xpreneurs, the LMU Entrepreneurship Center, the CDTM and Manage&More/UnternehmerTUM have already been proven as excellent sources for talent here at Kyte. If you are affiliated with one of these programs, we will almost certainly fast track your application and take a closer look.
- Munich is the heart of the German automotive industry. BMW is headquartered here, AUDI is just around the corner, and countless automotive suppliers are also around. The transportation infrastructure in Munich is one of the best systems in the world, and we like to be part of an ecosystem and environment that shapes the future of transportation.

Beyond that, there are certain personal touch points that made the decision ultimately very easy for us:

- Francesco and I met in our engineering studies at the Technische Universität München, we had our first so-called “company jam sessions” on mountain bike and climbing trips in the Bavarian Alps. We also received initial angel funding from business angels from Munich.
- In addition, California and Bavaria are naturally close neighbords in a lot of dimensions - mountains, climbing, surfing the Eisbach River Wave, … there are direct daily flights between San Francisco and Munich (not during Covid19), which will make the connection between the two offices even easier.

Ultimately, we believe Kyte can combine the strength of Silicon Valley’s venture capital, proximity to software technology decision makers and partners with the excellence and talent access of the Munich engineering ecosystem.

## How we are starting

For now we are operating 100% virtually in Munich given the Covid19 situation, but very soon we will be opening a physical location; we are already looking out for spots! I couldn’t be more excited to already have 3 strong engineering/product folks on the ground in Munich: Rafael, Pedro and G. (not announced yet). They are already rocking the show and are moving mountains.

![Kyte markets and offices. {caption: Kyte markets and offices.}](https://kyte-cms-media-prod.s3.amazonaws.com/blog_munich_expansion_map_2818f501d4.png)

## What the perks are for team members in our Munich office

We offer all of our Munich based engineers and product folks the opportunity to spend around 2 weeks a year in our San Francisco office and sponsor travel and accommodation. The goal here is to strengthen the Kyte culture, facilitate jam sessions and workshops, and make sure we work towards even better alignment across teams.

We invite all engineers in/around Munich to take a look at our openings, we are very excited to become a fundamental part of the growing and thriving Munich tech ecosystem. Check out our positions on our [Kyte Job Board](https://jobs.lever.co/kyte).

Kyte is coming home - to some degree. We are already looking for our first official Oktoberfest offsites (in a post-Covid world).

Cheers and Prost!

blog-munich-expansion-oktoberfest.jpg

Back to the roots: Kyte expands its engineering and product team in Munich, Germany

Once booked, your Kyte will be delivered directly to your doorstep by one of our “surfers” (= Kyte delivery driving partners) saving you the hassle of the traditional rental car pickup, waiting in line and getting frustrated by filling out a ton of paperwork. In this article we want to shed some light on one of the most central challenges that our engineering team is solving for: Automatically finding the best surfer to deliver your car exactly where and when you need it.

This article is the first one as part of a series of engineering-focused posts that are giving some background on what’s happening under the hood at Kyte, what engineering culture is like at Kyte and what other exciting problems we are tackling.

![San Francisco map](https://kyte-cms-media-prod.s3.amazonaws.com/blog_surfer_delivery_kepler2_7e5f3d69e7.png)

## Our Goal: Improving the Matching between Surfer and Kyte Trips Helps both Customers and Surfers

There are plenty of reasons why having a good automated matching algorithm is beneficial for both our customers and surfers:

- Customers get their Kytes as fast as possible: Within seconds after a booking comes into our system the algorithm can find the best surfer available that can start right away with preparing the vehicle and heading over to the delivery location.
- We can offer customers more flexibility to make changes to their booking: Using the Kyte app our customers can extend or change their drop-off location. The matching algorithm allows us to dynamically check which alternative drop-off locations and times we can offer based on how busy our surfers are during that time.
- We maximize our surfers' earnings by making their schedule more efficient: The matching algorithm optimizes the sequence of trips for each of our surfers. We minimize idle and transit times. This allows our surfers to do more trips and consequently earn more during the time they want to work.
- We can use our available Kyte fleet as efficiently as possible: With a growing number of trips we can exploit system efficiencies. For example if a vehicle is dropped off close to where a car will be picked up soon, our surfer can actually clean and prepare the vehicle in that are rather than bringing it all the way back to our depot.

## Problem statement: Surfer Matching is a Hard Mathematical Problem with Temporal and Spatial Constraints

Let’s take a closer look at the structure and constraints of the mathematical problem that lies behind finding the best surfer for a Kyte trip. In its most basic form (disregarding spatial constraints and assuming a fixed delivery and return time) the problem can be seen as a classical [interval scheduling](https://en.wikipedia.org/wiki/Interval_scheduling) problem. The only hard constraint would in this case be that a surfer can only do one trip at a time. The figure below shows an example setting where the green three trips need to be fulfilled by any of the three surfers.

![Diagram 1 {caption: Matching diagram 1.}](https://kyte-cms-media-prod.s3.amazonaws.com/blog_surfer_delivery_matching1_e5f893dcee.png)

This interval scheduling problem can be translated into a graph coloring problem. Where each trip or unavailability is a node in the graph and we draw edges between all the nodes that overlap on the time axis. Each color corresponds to a surfer. Finding a schedule that works is then equivalent to finding a graph coloring so that neighboring nodes never have the same color. While there are many efficient algorithms for coloring a completely uncolored graph, it is in general a very hard (NP-hard) problem to finish coloring a graph with pre-colored. Since surfers are not available all day we will always find pre-colored nodes in our graph (see the “Unavailable” nodes in the Figure below).

![Diagram 2 {caption: Matching diagram 2.}](https://kyte-cms-media-prod.s3.amazonaws.com/blog_surfer_delivery_matching2_1c6f4770ff.png)

On top of that, the graph representation can only contain temporal hard-constraints while in reality the time needed for a delivery or return highly depends on the surfers current location. This is why we need a further analysis of all the potential solutions we can find to the graph-coloring problem. And here is where heuristics for non-convex optimization problems come into play.

In the case of the situation in the figure below we ultimately have two different options for the return trips, it could either be done by Surfer 2 or Surfer 3.

![Diagram 3 {caption: Matching diagram 3.}](https://kyte-cms-media-prod.s3.amazonaws.com/blog_surfer_delivery_matching3_0ef5344149.png)

Which out of these options is the better one depends on multiple factors that we have not considered in our temporal hard constraints so far:

- Transit times between consecutive trips: If a surfer has a consecutive trip in a similar region it is beneficial to offer the trip to him. This will minimize the transit time.
- Possibility to do vehicle “flips”: Cars of the same category that are picked up can potentially directly be delivered to the next customer. However, we also need enough time between the trips to check, sanitize and refuel the car.
- Initial location of the Surfer: If the surfer does not have another trip before it is better for him to start with a trip close to where he currently is.

To distinguish the best option we designed a cost function that takes the aforementioned parameters into account. Unfortunately, this cost function does not fulfill the criteria for a convex function. This means that if we want to find the optimal match between trip and surfer given all of these criteria there is no algorithm that can guarantee to deliver the global optimum solution in finite time.

Fortunately, we do not need to have this guarantee but it is sufficient for us to find a sufficiently good (close to optimal) solution. There are multiple heuristics that find such a sufficiently good solution efficiently. In the next section we will talk a bit more about which of these heuristics we decided to use and how it concretely works.

## Approach: We use a Heuristic that Helps us Find a Nearly Optimal Solution in Reasonable Time

We are using a popular heuristic for non-convex optimization problems.

In general it is based on two core principles:

- Exploring neighborhoods to gradually find better solutions: We start with a random solution and then evaluate how the solution performs if we change it a little bit. From this “neighborhood” around the current solution we pick the one that performs best. Then we create a new neighborhood around this solution and again pick the new best, and so on. This means that with every step we take we will gradually improve until we have a sufficiently good solution.
- Keeping a list to prevent being stuck: There are cases where strictly adhering to the principle above leads to a behavior called cycling. It means that we always go back and forth between two solutions that are okay but we never make steps far enough to actually explore a region of solutions that are even better. To prevent this we keep a list of all the solutions we already picked in the past. We will then never go back to a solution that we already visited recently. This forces the algorithm to explore sub-optimal directions from time to time which prevents being stuck in a local optimum.

## We have ambitious plans! - Join us to make them reality!

While we are already using the first version of the matching algorithm in our day-to-day operations we don’t stop here. We have ambitious goals to add a plethora of features and make our algorithm even smarter. Let us give you a quick look into some things we are currently working on:

- Optimize short notice bookings in batches: As we get more and more short notice bookings in, it will be beneficial to group them in batches and then optimize for this group of new trips. This will make the overall schedule better compared to taking a greedy approach for each single trip as they come in.
- Use machine learning to better predict transit times in between trips: Since we explore hundreds of thousands of potential schedules it is infeasible to compute the exact transit estimates for each of these options. However, we can learn a heuristics that will give us a sufficiently good and fast estimate of how much gap in between trips to plan for during a certain time of the day and depending on both trip locations.
- Dynamic Pricing for customers: Sometimes it doesn't really matter to customers if they drop off their Kyte at 8 PM or 9 PM but depending on the availability of surfers it might be much easier for them to accommodate for one of these options.

As you can see we have lots of exciting projects on our agenda! If you are somebody that loves to work on real-world hard engineering challenges that combine the physical and digital world in a fast-paced environment, we would love to get in touch with you! Please don’t hesitate to shoot us a message at careers@drivekyte.com and we can give you some more insights on what we are working on!

blog-surfer-delivery-kepler1.png

Engineering Insights: How we use Non-Convex Optimization to find the Best Surfer to Deliver your Kyte

Kyte, the leading door-to-door, app-based car rental delivery company, has officially launched in the greater Los Angeles area, expanding its service area in California after launching in San Francisco last year. Kyte is experiencing up to 40% MoM growth. With COVID-19 in the backdrop, Kyte has seen its demand increase exponentially because of its contactless, mobile app-first service.

Kyte’s rental cars are delivered and picked up at any address in the service area, which includes cities such as Downtown Los Angeles, El Segundo, Santa Monica, Westwood, Culver City, Marina Del Rey, Manhattan Beach, Inglewood, and West Hollywood. Users can use promo code _KYTELA_ and receive a discount on the first ride.

“Kyte’s goal is to free people from car ownership and to transform the otherwise unchanged car rental industry,” said Kyte co-founder Ludwig Schönack. “Angelenos will now have access to a car with a push of a button, while avoiding long lines.”

The biggest complaints with the car rental experience have been the required travel to and from centralized pick-up and drop off locations, the long wait in lines and the lack of fare transparency. Kyte has solved this challenge by making its technology a mobile app-first experience. Users can book their rental on Kyte’s app or through its online portal, [kyte.com](https://kyte.com), and then the company dispatches a “Surfer” to deliver a car within two hours. Users are charged a daily rate; long-term discounts are automatically applied for frequent users.

Kyte is not a car-sharing or peer to peer rental platform. Their rental cars are sourced from local professional fleet companies, and are sanitized before deliveries to ensure users’ safety. Kyte has built strategic partnerships with large fleet owners and rental car companies to deploy the professionally managed fleets to users through Kyte’s proprietary tech platform. The company aims to maximize the utility of fewer vehicles and significantly reduce the unnecessary and excessive manufacturing footprint that stems from overfleeting and limited utilization.

Since Kyte’s soft launch earlier in the year, fueled by its zero-asset model, it has already become one of the fastest-growing players in transportation. Today, the company has offices in Munich and San Francisco with a growing team that held key positions at companies like Uber, Chariot, Lime, and many others that helped to shape the transportation ecosystem. It plans to launch in metropolitan areas such as Boston, Washington D. C., Seattle, and six other cities by 2021.

## About Kyte

Kyte aims to replace car ownership by offering access to cars on demand. By partnering with large professional fleet owners, Kyte delivers a reliable, mobile-first, hassle-free alternative to car ownership. With key investors and stakeholders from the travel and transportation tech space, along with an ambitious founding team and a robust network of partners in the automotive and rental space, Kyte strives to dramatically decrease car ownership and increase short-term car access for all. Kyte is headquartered in San Francisco with offices in Munich.

Learn more at [kyte.com](https://kyte.com) 💚

blog-kyte-expands-to-la.jpeg

Kyte Expands to Los Angeles

Since February, our prevailing message was “If you can stay home, please do so." However, in the past weeks since San Francisco has broadly gone into shelter in place , there’s been one group of workers that have been much discussed in the media, for which that is simply not an option - the essential workers.

“There isn’t a tele-burger, and there isn't a tele-Amazon either.”

While many firms, like Apple, Microsoft and Google, are enthusiastically following the advice, for more than 70% of workers, working from home is not an option. Only about 42 million of 144 million U.S. workers could do their jobs at home in 2018, according to the U.S. Bureau of Labor Statistics. Millions of other Americans in fields like retail, transportation, logistics, food and other industries, can’t simply log on to Slack or Zoom to do her job and for those safe and reliable transportation options from the public and private sectors become increasingly scarce.

Just as public transit authorities in San Francisco, the LA Metro, and private transportation providers are forced to continue to scale back services amid the outbreak, Kyte started operating in Los Angeles to help. We’re proud to work with some of the largest hospitals and essential businesses in San Francisco and LA and offer safe and reliable alternative transportation during these times.

As always, we will continue to deliver our cars straight to the door. Now more than ever, our service is the best way to rent a car given that we can avoid all human interaction and unconditionally comply with social distancing measures. This means you receive a sanitized car at your desired address, which helps you to avoid shared transportation, crowded rental counter queues and unnecessary travel in the process.

What we’re doing in LA:

We’re discounting long term bookings up to 45% for immediate short term and for the month of April, we’re offering an additional LA-specific discount of 10% with the discount code _LABETA10_.

We’re building custom offers for any essential businesses to get their teams to where they’re needed most. Please reach out to support@drivekyte.com to talk to an agent
Please continue to follow the government’s advice on COVID-19.

#staysafe #drivekyte 💚


blog-kyte-launch-la.jpeg

Why Kyte beta-launched in LA amidst Covid-19 outbreak

## Getting bored due to the lockdown? Go for a hike (Yes it's allowed and even encouraged!)

As millions of us remain confined in our homes due to coronavirus, there has been [quite a broad confusion](https://edition.cnn.com/travel/article/hiking-beach-coronavirus/index.html) on whether it’s permitted to go for a walk and get some fresh air (and much needed exercise).

The recent lockdown in the Bay Area has closed down most centers for recreational activities, including fitness centers, gyms, zoos, and campgrounds. But not all hope is lost - you are still allowed (and encouraged) to go out for a hike away from the crowds.

[Many authorities and experts](https://www.timesheraldonline.com/2020/03/19/coronavirus-is-it-legal-to-go-hiking), in fact, recommend outdoor activity — with caution. Whether it’s for a quick exercise in the backyard or a hike in secluded areas, being momentarily outside helps reduce staying-at-home frustrations. Read on for some general advice about going outside during the pandemic, and make sure to keep tabs on local guidance as events evolve.

## How can you avoid crowds?

Preferably it's best to reach parks which are some distance away from downtown. At the same time, it’s advisable to **avoid public transport or shared transport**. The best course of action is to use your own private car. San Francisco tech based company, Kyte, offers delivered sanitized cars to any address in the city. You can rent a car for a day, weekend or even for weeks and months. Once you are done with your adventure, [Kyte](https://kyte.com) will pick up the car from your desired address.

blog-bored-during-lockdown-hike.jpeg

Getting bored due to the lockdown?

With recent news around coronavirus (COVID-19), we wanted to share the actions we're taking to continue making Kyte a safe transport option.

### What Kyte is doing
- Sanitizing every car before it is getting delivered. We are sanitizing all surfaces customers might touch before every delivery. This makes the car a neutral zone as you enter it.
- Decreasing contact during handovers. Every surfer has been provided with sanitary gloves to make the handovers of your car touch-free. Also, we have updated our handover process to allow customers to sign their rental agreement on their own phones. This is for both your and our surfers’ safety.
- Introducing flexible long-term rentals following increased requests from customers who want to avoid other means of transport.

### Things you can do
- Stay informed using official sources. The CDC and World Health Organisation (WHO) are constantly updating their pages with the latest information about the coronavirus.
- Practice good hygiene. Clean your hands often, and use a hand sanitizer that contains at least 60% alcohol. Cover your mouth and nose when sneezing, and avoid touching your eyes, nose, and mouth with unwashed hands.
- Stay at home if you are sick. If you are feeling unwell, please avoid using rideshare, car rentals, taxis, or public transportation.

Stay safe! - Team Kyte.

blog-your-safety-sanitize-car.jpeg

We care about your safety

In the recent events around coronavirus, we are launching heavily discounted long-term rentals to provide everyone with safe, clean and reliable means of transportation during this period of crisis.

In the past week, we discovered two things: Commuters are looking for an alternative to public transport and shared ride hailing. Kyte was mentioned in this Mashable article as a way to avoid these forms of transport. In general, the public needs a reliable way of safe, private transportation to get out of the city in case the situation gets worse. This is even highlighted by the fact that all rental car businesses have been deemed as "essential" (by the city) and thus are exempt from upcoming business lockdowns.

In addition to the extra safety and cleanliness measures that we implemented last week, we see a certain responsibility in reacting to the crisis and doing what we are good at: rapidly building and shipping meaningful products that matter to people around us!

Huge kudos to the team for making this happen so quickly! Our supply team has been working hard on expanding our fleet and accessing affordable vehicles. Our product and engineering teams are working hard on integrating this product offering into our mobile and web product flow, and our operations team is getting these delivered to you now!

If you need a clean, reliable car for weeks or months, please take a look at our long-term rental discounts here: App or web. Contact us with any questions that you might have!