Founder
Artur Czemiel
Founder of Aexol AI
wrote the tooling a lot of GraphQL teams still use. Now he's building the spec language that tells AI coding agents what to build.
Aexol AI is a spec-driven agentic development environment. Before you trust a product, look at who writes the language. Fifteen years public on GitHub, projects he founded at 6,065 and 1,976 stars, and a founder-owned npm scope shipping Aexol. The receipts are below — every one of them is a link.
Counts as of October 2026.
The arc
Fifteen years, in public
Earlier
VFX, Python, and a patent
Writes Python scripts for movie visual effects, patents an algorithm for weighing small objects with a smartphone, and holds a computer-science engineering degree (thesis: measuring temperature with a smartphone and sound waves).
2011
The company and the account
The founder's GitHub account is opened in May 2011; the earliest public repositories follow in 2012. According to aexol.com, the company has been shipping client software since 2011 as a 100% senior team.
2018
GraphQL Editor
Launches GraphQL Editor, a visual node editor for GraphQL — now 6,065 stars, MIT-licensed. The Hacker News launch thread collects 205 points and 23 comments. The first @aexol package lands on npm in February.
2019
GraphQL Zeus
Publishes GraphQL Zeus: read a GraphQL schema, emit fully typed clients you compile against. 1,976 stars and still maintained in 2026. The graphql-editor org grows to roughly 45 repositories.
2023
Axolotl
@aexol/axolotl appears on npm in October — a type-safe GraphQL backend framework (89 versions published, 27★ in the 53-repo aexol-studio org). Aexol AI is built on top of it.
Nov 2025
Aexol AI
Starts building Aexol.ai: the spec-driven agentic development environment at aexol.ai — the AEXOL stack of SDL, Studio, and Spectral.
2026
Spectral ships, at cadence
@aexol/spectral's first npm publish is May 2026. By October 2026 it has 301 published versions (latest v0.9.249) and roughly 28,500 downloads a month.
Proof, not promises
on GraphQL Editor, which he co-founded. every one of them public. this page deals in entries you can open in another tab — check all of them, that is the point.
Every claim in this table is a link
Marketing pages deal in adjectives. This one deals in entries you can open in another tab. Check all of them — that is the point.
- Fifteen years public on GitHub
- GitHub account opened May 2011, earliest repositories from 2012 — 70 public repos, 114 followers. github.com/aexol ↗
- 6,065 stars on the GraphQL editor he co-founded
- A visual GraphQL IDE, MIT-licensed, under the graphql-editor org (~45 repositories, created 2018). github.com/graphql-editor/graphql-editor ↗
- A typed-client generator still maintained in 2026
- GraphQL Zeus: 1,976 stars, created 2019, homepage zeus.aexol.com — schema in, compilable typed clients out. github.com/graphql-editor/graphql-zeus ↗
- The founder owns the entire @aexol npm scope
- Publisher arturczemiel owns roughly 46 @aexol packages; the first was published in February 2018. npmjs.com/~arturczemiel ↗
- Release cadence on the product you are evaluating
- @aexol/spectral: first publish May 2026, 301 published versions since (latest v0.9.249, Oct 2026), ~28,500 downloads per month. npmjs.com/package/@aexol/spectral ↗
- A Hacker News launch that landed
- “GraphQL editor — Visual node editor for GraphQL”: 205 points, 23 comments (2018). HN user aexol, 269 karma. news.ycombinator.com/item?id=18045532 ↗
- Wrote the book. Literally.
- “GraphQL Best Practices”, published by Packt. GraphQL Best Practices on Amazon ↗
- A registered company, not a landing page
- Aexol is a Polish AI-native software house (Białystok, KRS 0000602817). In its own words: “Since 2011 we've been shipping software… 100% senior team.” aexol.com ↗
Why this person builds this
Look at the portfolio and one thing repeats:
take a schema — a precise, formal description of what a system is — and make it produce code.
GraphQL Editor treats the schema as a visual, manipulable source of truth. Zeus reads a schema and emits typed clients you compile against; when the schema changes, your build breaks until you regenerate. Axolotl runs a schema end-to-end into a fully-typed backend.
That is not prompt-and-pray. It is schema-first, compiler-checked tooling, and it took roughly a decade of public work to build.
So the Aexol thesis — write an exact specification of what you want and let an agent generate the implementation — is not a trend he joined. It is the same instinct, aimed at a new kind of code generator. The difference is only who reads the spec now: an agent instead of a code generator. The Aexol language and its tooling come from the same person and team who wrote every receipt on this page.
The lineage, delivered
What fifteen years of tooling discipline buys you
A language that can be parsed and validated
The stack includes a zero-dependency parser for the Aexol language with semantic validation — the spec is checked before anything generates, not after. The same team that shipped Zeus and Axolotl built this language and its tooling.
Generation that has to typecheck
Zeus has been emitting compilable, typed clients from schemas since 2019; Axolotl has been a fully-typed backend framework on npm since October 2023. Aexol AI's generated artifacts — GraphQL schema, Prisma models, API routes — sit in that same lineage.
A release cadence you can read
No promises about shipping fast — the npm registry lists 301 published versions of @aexol/spectral between May and October 2026, and the changelog on this site dates the releases.
Software-house delivery standards
The company behind the product is a registered Polish company (KRS 0000602817) that, per aexol.com, has shipped client software since 2011 with a 100% senior team.
Written work
He has written this argument down before
The book is the long version of the same idea: schemas and specifications deserve to be taken seriously. He also publishes as aexol on dev.to (profile: “CTO at GraphQLEditor, Aexol”, joined 2019).
FAQ
Questions a skeptical engineer would ask
01Is this a side project of a random startup?+−
The company is a registered Polish software house (KRS 0000602817) that, according to aexol.com, has been shipping client software since 2011 as a 100% senior team. The founder's GitHub account opened in May 2011 and carries 70 public repositories; the projects he founded carry 6,065 stars (GraphQL Editor) and 1,976 (GraphQL Zeus) under the graphql-editor org. He has been building Aexol.ai since November 2025.
02Who actually maintains the npm packages?+−
The founder's npm account, arturczemiel, owns the entire @aexol scope — roughly 46 packages, published since February 2018 — plus graphql-zeus (unscoped; July 2019, still on 7.x: latest 7.2.1, April 2026), @aexol/axolotl (October 2023, 89 versions), and @aexol/spectral (May 2026, 301 versions). The full history is verifiable on npm itself.
03How long has this actually existed?+−
The product at aexol.ai has been in development since November 2025, and @aexol/spectral has been on npm since May 2026. The approach behind it is older and public: schema-first tooling with generated, type-checked code — GraphQL Editor (2018), GraphQL Zeus (2019), Axolotl (2023).
04Can I audit the code?+−
Partly, and we will be straight about it: the Aexol language core and the platform are not on public GitHub. What is public: the @aexol packages on npm, the changelog on this site, and the older open-source work — GraphQL Editor, Zeus, Axolotl — which is the best evidence of the engineering standard this team holds itself to.
05“Spec-driven AI” sounds like a buzzword. What is actually different?+−
It is a mechanism, not a slogan: a .aexol spec is parsed and validated by a zero-dependency parser before anything is generated, and Studio shows the artifact set the pipeline will emit ahead of the run. If you ever used Zeus — schema in, typed code out, compiler checks it — you have already seen this idea working, years before AI agents were the audience.
Judge the output
See the output before the agent runs
Write a .aexol spec, validate it, and look at exactly which artifacts the pipeline will emit — ahead of generation. That is the fastest way to judge whether this founder's standards are in the product.