EN
Back to the archive

The archive · Developer & Business Tools · Product decision · 2024–2025

Tessl raises $125M to bet AI-native 'spec-centric' development replaces code-centric tools

Snyk founder Guy Podjarny's London startup raises $125M to build a spec-centric platform where humans specify and AI writes, debugs and maintains software.

Tessl

The betThat software creation should flip from code-centric to spec-centric: humans specify what they want, AI implements and maintains it — the spec is the source of truth.Live

What the business is

A London-based developer platform for what Podjarny calls AI-native software development: developers express intent in specifications, and AI writes, tests and maintains the implementation.

Starting capital$125M total: a $25M seed led by boldstart ventures and GV, then a $100M Series A led by Index Ventures with Accel, GV and boldstart; reported $750M valuation

How it started

Tessl was founded in London in early 2024 by Guy Podjarny, who previously founded Snyk and served as Akamai's CTO after it acquired his first startup Blaze.io. In April 2024 it quietly raised a $25M seed round led by boldstart ventures and GV, then stayed in stealth while building.

What happened

On 14 November 2024 Tessl came out of stealth, announcing a $100M Series A led by Index Ventures with Accel, GV and boldstart, at a reported valuation of $750M. At the time it had about 21 employees and two preliminary product versions in internal testing, and opened a waitlist for the platform ahead of an early-2025 launch.

How it ended up

In September 2025 Tessl shipped its first products: the Tessl Spec Registry, an open-beta catalogue of over 10,000 specs for external libraries, and the Tessl Framework, a closed-beta platform that uses specs and tests to keep AI agents from hallucinating APIs and breaking existing functionality.

Background

Tessl is the third startup from Guy Podjarny, the founder of Snyk (developer security, later valued in the billions) and a former CTO of Akamai. His bet, laid out in November 2024 when Tessl emerged from stealth with $125M, was that AI coding assistants were too incremental: they automate the existing code-centric workflow instead of replacing it.

Tessl's answer is 'AI-native software development' — a spec-centric platform where developers express what they want and AI implements, tests, debugs and maintains it. The specification becomes the source of truth, freeing the developer to act like an architect while the platform handles maintenance, security fixes and dependency updates.

The money came in two rounds announced together: a $25M seed from boldstart and GV in April 2024, and a $100M Series A led by Index Ventures with Accel, GV and boldstart, at a reported $750M valuation. The company then ran a long waitlist campaign before its platform launch in early 2025.

A year later, in September 2025, Tessl shipped its first real products: the Spec Registry, an open-beta library of more than 10,000 specs that stop agents hallucinating APIs, and the Tessl Framework, a closed-beta platform that uses specs and tests as guardrails. The bet had moved from slides to software, with adoption and revenue still unproven.

What has to be true

  • Podjarny's track record (Snyk, Akamai) let investors fund a first-principles redesign of software development rather than another copilot feature.
  • The thesis targeted a real pain point: agents hallucinate APIs and break code, so Tessl made specs and tests the guardrails for agent reliability.
  • Announcing the seed and Series A together created one big moment that put the reported $750M valuation into the news cycle.
  • Spec-centric development is a category bet: if it works, Tessl owns the system of record for how software gets made; if not, it is a very well-funded dev tool.

What can be applied

A credible founder can sell a paradigm shift: Podjarny raised $125M on the thesis that AI assistants were incremental and the real value was rebuilding the platform around specs, not code.

Aftermath

As of 16 September 2025, Tessl had launched its first products — the Tessl Spec Registry (open beta, with more than 10,000 specs) and the Tessl Framework (closed beta) — both aimed at making AI agents reliable in professional codebases. The company had not disclosed revenue or customer numbers, and the open platform promised at the Series A had been reshaped into these two products.

Sources

spotted an error? The archive wants to know.

Your turn

You just read one. Describe what you are building, and see who is betting on the same thing.

Free account · 3 free questions · no card

Related cases