Announcing Our Series A for AI Native Software Development

Today is an exciting day at Tessl: we’re announcing $125M in funding, including an April $25M seed round led by boldstart and GV, and a new $100M Series A led by Index with participation from Accel. We’re happy to be well-fueled for the big journey ahead, and humbled by this vote of confidence in our mission and team.

Software powers our digital world, and improving the tools we use to create it can have exponential effects. AI dev assistants offer immediate productivity gains, so it’s no surprise that they’re getting widely adopted. However, at Tessl, we look to a future beyond these incremental improvements. We believe unlocking AI’s true potential for developers means rethinking software creation from first principles.

From Code-Centric to Spec-Centric

Right now software development revolves around code – writing, editing, reviewing, testing, merging, packaging, versioning, and observing… code. Code is the source of truth for the application, and it couples what the app does with how it does it, mixed into the very same lines and increasingly entangled as the application matures. Knowing what the app should do requires user and business knowledge, while implementing the how requires technical coding skills. This coupling makes apps slow and expensive to create, as well as risky and fragile to change. It also means maintenance is a never ending grind, often taking up the majority of an engineering org’s time. 

AI offers a breakthrough: finally separating what the software does from how it does it. Large Language Models (LLMs) let us explain requirements in natural language, making it easier than ever to describe what you want to build and filling in gaps as needed. On top of that, they enable machines to handle how to implement those ideas in code, complete with an understanding of both the technical context and the intent behind new features. By combining this ease of specification (‘spec’) with AI-powered implementation, we can imagine a new way to create software.

Introducing AI Native Software Development

We call this new development paradigm AI Native Software Development, and it is spec-centric, as opposed to code-centric. 

Developers will specify what they want, captured in a specification format that fits their needs, and AI will break down and implement the solution. The spec will be structured to enable bigger apps than today’s prompt-based creators, and evolve as the application matures. Creating software this way will be dramatically faster, and accessible to many more creators. When we lower these barriers, what more can we build?

Of course, the initial creation is only a small part of a software’s lifecycle. Once created, an AI Native Dev platform will continuously maintain the app, relying on tests from the spec to ensure the app doesn’t break and expanding those regularly. Dependency updates, known vulnerability reports, API changes and more will be autonomously and safely applied. How much time do you spend maintaining old software today? What else could you build if you got that time back?

All the optimizations that dev teams want to apply, but lack the time or expertise to do, can be captured in this machine-driven development pipeline. Security flaws will be found and fixed, along with accessibility issues, performance gaps and more. Docs will be updated on every change, compliance tests performed with proper attestations, and tests regularly updated. Over time, insights from telemetry will improve implementations, creating a tight feedback loop with little human effort. How much faster, safer and more reliable would the digital world be with these changes? 

AI Native dev promises to deliver such optimal flows, but also opens new doors we couldn’t consider before. Spec-driven apps will be language agnostic, easily adapted from JavaScript to Python, iPhone to Android or browser to native app. Implementations will be customizable, refined to your specific environment based on your data, and enabling dynamic changes like prioritized speed at peak hours vs costs overnight. How can we best leverage such elastic and personalized software? 

{{waitlist-launch}}

Join the AI Native Dev Movement

AI Native Software Development is just at the edge of the possible today. Some aspects of it can be done already, and just need better reliability and structure. Others might not be feasible, but are just around the corner. This convergence makes now the perfect time to start pioneering a new approach. 

Creating AI Native software requires an entirely different workflow to that of its code-centric predecessor. It includes steps like: writing and editing a spec; understanding and interacting with decisions the LLM has made; reviewing and troubleshooting implementation steps by guiding the LLM; versioning adaptable software; autonomous maintenance workflows; etc. Enabling such creation calls for a new development platform, which will form and evolve alongside the paradigm. This is what we’re building at Tessl. 

However, AI Native Software Development is bigger than Tessl. 

When a new paradigm is born, it can start closed or open. We have seen the power of open ecosystems in our careers, and fundamentally believe that the new AI Native software development ecosystem has to be an open and composable one. 

It must enable innovative tool builders, who will create building blocks and pathways that are awesome in different ways. It should empower developers, teams and companies to choose their special recipes for building software, and share successes, failures and learnings with the community. It must tap into today’s developers hard earned software crafting skills, despite them not writing code, and allow vastly different opinionated approaches to thrive. 

We’re committed to serving this community by facilitating conversations about what AI Native Software Development should be. Throughout the year, we’ve been finding and sharing excellent perspectives, tools and stories touching AI Dev, and will continue to do so – including at our AI Native Dev Con on November 21st.

Our open platform will be available in the new year, and we welcome early adopters to join the waitlist to be first to try it and provide feedback. Until then, we encourage all of you to join the conversation, follow its developments across the community, and help shape this new approach to software creation.