↓ Skip to main content
tomrochette.com
  1. Agents/

Tessl

Author
glm-5.3, glm-5.3-flash
History / Edit / PDF / EPUB / MD
Created: / Updated: / Status: finished / Readability: 3 / 4 minutes read (~734 words)
agent-curated 173 fully-ai-generated 263 llm=glm-5.3 170 llm=glm-5.3-flash 178 platform 1 research-note 148 spec-driven-development 5 venture-backed 1
Table of Contents

Tessl is Guy Podjarny’s (Snyk founder) spec-centric AI-native development platform: a web-first product where specs drive AI-generated (and human-written) software, funded by a $125M Series A and extended in 2026 with a skills package manager. Facts below verified as of 2026-09-13.

Tessl is the bet that spec-driven development is a platform business rather than a CLI convention, and its $125M raise, while the open tools give the method away free, is the category’s purest experiment in whether specs become infrastructure you rent.

What it is
#

A hosted platform positioned around zero-maintenance, spec-centric development: you express intent as specs, the platform manages the loop around them, and since January 2026 Skills on Tessl treats agent skills as software with a lifecycle and a package manager. Its open-source surface is deliberately thin: the tesslio/cli helper (70 stars) exists mainly as an enablement tool, and the substance lives on tessl.io.

Status
#

Funded and building, quiet on GitHub. The Series A announcement (November 2024) declared $125M for the platform; the founding announcement (October 2024) frames Podjarny’s pivot from Snyk’s security mission to AI-native development. The strongest third-party signal is Martin Fowler’s October 2025 analysis naming Tessl one of the three SDD pillars alongside Kiro and Spec Kit (the 128-point thread), which treats it as a serious approach while noting it was then the least mature of the three. The CLI has not been pushed since 2026-03-05, which is consistent with a web-first roadmap and worth re-checking before adopting (re-confirmed unchanged, now at 70 stars, on 2026-09-13).

Strengths
#

  • Founder pedigree and $125M mean the platform has runway the community tools will never have.
  • Skills-as-software with lifecycle management is ahead of the ecosystem and pairs naturally with the Skills category’s direction.
  • Being a platform, it can enforce the review gates that file-based conventions can only suggest.
  • The Fowler analysis treats its spec-centric model as coherent, not a fad.

Cautions
#

  • Web-first means your specs live in someone else’s platform; export and portability are the questions to ask before committing.
  • Venture-scale expectations in a category where the free tools (spec-kit 136k stars, OpenSpec 68k, BMad 53k) set the price anchor at zero.
  • The Vibe Kanban shutdown record in this section shows what happens to agent-layer SaaS without a business model; Tessl’s counter-bet is that specs are stickier than boards.
  • Third-party coverage beyond the Fowler piece is thin, and its GitHub surface is too small to audit.

Pricing
#

Platform tiers are now published on tessl.io (free since 2026-09-08, re-verified 2026-09-13): Free at $0 per month with 1,000 credits, Team at $100 per month with 5x the credits, and Enterprise custom. The credit model is subscription metering, not open source, as the raise implied.

Compared to
#

  • GitHub Spec Kit and OpenSpec: the free, repo-native toolkits; choose them when the specs must stay yours.
  • Kiro: the other closed, paid spec-first bet, an IDE rather than a platform; Kiro gates credits, Tessl gates the workflow.
  • BMad Method: the free full method; choose it when you want the process without the platform.

Bottom line
#

Recommended only for teams that evaluate it against a written exit plan: export what, portability where, cost at scale. Not for anyone whose spec ledger is a long-lived asset they intend to own. The disagreeable claim I will defend: Tessl will either prove specs are buyable infrastructure or become this category’s cautionary funding tale, and by its own metrics (open-source footprint near zero) it has chosen the riskiest possible position to prove it from.

Changes
#

  • 2026-08-27 - Created in the Spec-driven development expansion, recording the $125M Series A, Skills on Tessl, and the Fowler analysis; pricing left unverified.
  • 2026-09-08 - Recorded the first published platform pricing and rewrote the Pricing section (Free, Team $100 per month, Enterprise).
  • 2026-09-12 - Recorded the Tessl Code Review launch, with pricing tiers unchanged.

See also
#

References
#