Early access — first draft
APIs.io On-Premise
The same machinery that finds, profiles, scores and publishes 142,000+ public APIs from 27,600+ providers on APIs.io — running inside your walls, over your APIs, on a model you host yourself. Nothing leaves your network.
Request a quote See how the Kin Score works
The problem
Most organizations cannot answer a simple question: what APIs do we have? The answer is scattered across hundreds of Git repositories, Confluence spaces, old developer portals, API gateways and docs sites — some with an OpenAPI, many with nothing but a wiki page. Your teams, and now your AI agents, can’t reuse what they can’t find, and you can’t govern what you can’t see.
That is exactly the problem APIs.io solves for the public web. APIs.io On-Premise solves it for your organization.
What it is
A self-contained appliance that starts on a single Mac mini (Mac Studio next), running
gpt-oss-120b — OpenAI’s open-weight model, in MXFP4 quantization (a 59 GB GGUF) — served
locally by llama-server. The model does the reading, profiling and writing that the public
APIs.io pipeline does, without a token of your source code or documentation going to a hosted AI
provider.
It then adapts to how you already work, instead of asking you to adopt a new platform:
GitHub, GitLab, Bitbucket
Crawls the repositories you already have and manages every artifact it produces as Git repositories — the same way APIs.io manages public APIs today.
Claude, Copilot
Fits alongside the AI assistants your teams already use, so the catalog it builds is something those tools can read and act on.
GitHub Pages, AWS, Azure, or any static host
Publishes your own private APIs.io as a static site, wherever you already host internal sites.
How it works
-
1
Set up the appliance
We stand up the box on your network, connect it to your Git platform, and agree on what it is allowed to read.
-
2
Spider for APIs
It crawls your existing Git repositories, Confluence, internal portals and documentation, looking for APIs — OpenAPI, AsyncAPI, GraphQL, Postman collections, MCP servers, and the APIs that are only described in prose.
-
3
Aggregate into one place
Everything it finds is pulled together under a single GitHub organization (or GitLab group, or Bitbucket workspace) — one repository per API producer, with an APIs.json index and its contracts, schemas, examples and rules. See a public example.
-
4
Profile and score
Each API is profiled and scored with the Kin Score, re-weighted to your priorities — your governance rules, your security posture, your agent-readiness goals — with a blueprint of what each team can do to raise it.
-
5
Publish your own APIs.io
It generates and publishes a searchable APIs.io for your organization — provider pages, API pages, scores, and the artifacts behind them. See what a provider page looks like on the public catalog.
-
6
Iterate and adapt
The profiler, the publishing and the scoring evolve to the way your organization operates. It keeps crawling, so the catalog stays current as teams ship.
Why on-premise
- Your code and docs stay inside. The model runs on hardware you control. No API keys to a hosted AI provider, and no per-token bill.
- It is built on what you already run. Git is the system of record, a static site is the front door. There is no new database or SaaS platform to approve.
- It is proven on the open web first. Every step above runs every day across the public APIs.io catalog. You get the same approach, pointed inward.
Request a quote
Tell us a little about your environment and we will reply with next steps and a quote. Every request is read by a person.
Rather talk first? Email info@apis.io.