Snyk founder Guy Podjarny's Tessl raises $125m across seed and Index-led Series A for spec-first coding

Tessl, the London startup set up this year by Snyk founder Guy Podjarny, has raised $125m across two rounds it had not disclosed until now. The total combines a $25m seed from April, led by boldstart ventures and GV, with a $100m Series A led by Index Ventures, in which Accel, GV and boldstart also took part.

The size of the cheque stands out because there is nothing on sale yet. Tessl had 21 staff at the time of the announcement, two early builds of its software under test and a waitlist in place of customers. People familiar with the deal put the Series A valuation at about $750m, a figure the company has not confirmed.

A bet on the founder as much as the product

Investors are backing a track record. Podjarny's first company, Blaze.io, worked on making web applications load faster and was bought by Akamai, where he went on to serve as chief technology officer. He then founded Snyk, which scans code for security flaws and suggests fixes.

This is his third start-up, and he describes it as the most ambitious of the three. His earlier companies made software quicker and safer, he said; this one sets out to change how software gets written in the first place.

Index partner Carlos Gonzalez-Cadenas said his firm had known Podjarny for years. He summed up the investment thesis in one line:

“Software ate the world. Tessl is opening up the kitchen.”

For European investors the round is also a sign that the region can hold on to repeat founders in the AI tooling race rather than lose them to San Francisco. Tessl is headquartered in London.

From writing code to writing specs

The core idea is to move the developer's main job away from code and towards a specification. Under Tessl's model, an engineer states what a program should do and what counts as correct behaviour, and AI models produce the implementation. The same models are meant to take on the follow-up work too: debugging, security testing and keeping the code up to date over time.

Podjarny gave an example of what that could mean. A developer might tell a service to favour low latency at busy times of day and to favour lower cost when traffic is light. In principle, the system could then run quite different code, even in different programming languages, at different hours, with the spec as the fixed reference point.

That puts the developer closer to a systems architect or product manager, Podjarny said, spending time on user needs, problem-solving and design rather than on line-by-line implementation. Humans stay in charge of decisions, according to the company; the machine handles the typing.

He was blunt about the tools already on the market. "These AI dev tools are very much focused on automating the existing workflow," Podjarny said, describing them as built around code rather than intent.

A crowded field of coding assistants

Tessl is entering a busy market. GitHub, owned by Microsoft, sells Copilot, which completes code inside the editor and has a large user base. Cognition's Devin aims to build whole programs from plain-language prompts, and Codeium, Replit and Tabnine are among a long list of other companies offering AI help to programmers.

The startup argues that these rivals speed up today's workflow without questioning it. Whether developers want to give up direct control of their code is the harder question. Specs written in natural language can be vague, and teams that answer for production systems tend to be cautious about code they did not write and cannot easily read.

Tessl says the way to win trust is through community as well as product. Alongside the platform, it is building a group of developers who build with AI from the start, and it runs the AI Native Dev podcast and an AI Native DevCon conference.

Where the $125m goes

The company said the money will pay for building out the platform and for hiring across AI research, engineering, product, design and developer relations. With 21 people on the payroll, that leaves a long runway for a team still finding product fit.

The waitlist opened on the day of the announcement, and Tessl said it aimed to launch the platform in early 2025. Until then, its two internal versions stay with the team and a small group of outside testers.

Since the round, the company's pitch has moved towards software factories, built from versioned context, specs and agent skills that AI coding agents can draw on, plus repeatable agent workflows intended to produce reliable code at scale. The spec-first idea from the funding announcement still sits at the centre of that offer. The test for Tessl is whether engineering teams paying for AI agents will adopt it as the layer that keeps those agents on track.