A retrieval layer you embed —
not a vendor you wire into.
If you're shipping AI on top of someone's data, you've felt the trap: every source is a different OAuth, a different search API, a different permission model. inseam is the layer that flattens that into one contract — and gives you scoped, multi-party retrieval your product can sell against.
Things you can build on inseam
you can't build on anything else.
Not "done better." Done at all. Three concrete shapes that fall out of the Connection / Source / Access split — and out of a plugin contract small enough for an agent to fill in.
A customer signs in. Your AI answers about their emails, their invoices, their account — and can't see anyone else's, by how the index is built. The data envelope (From/To, account ids, party fields) becomes the identity signal; Access turns that signal into a scoped read path, evaluated at fetch time. Today this is built bespoke per company — or, far more often, not at all, and the chatbot stays a useless toy.
Tell Claude: "Build an inseam plugin for AcmeFlow." It invokes the
plugin Skill, scaffolds a ConnectionDefinition, registers a
LocationKind, writes the typed Source-write code, generates
fixtures, runs the framework-mandated test suite until green. Twenty minutes
later you inseam plugin install ./acmeflow and the niche tool
your team depends on is indexed alongside Gmail and Drive.
Any agent — your local LLM, ChatGPT, Claude, a custom Worker — talks to an inseam node the same way. Protocol skins are implementation details. You stop re-integrating per agent vendor; the node federates across the user's other nodes transparently.
Why the contract is small.
The whole plugin surface is three things: a ConnectionDefinition (how
this provider authenticates), a LocationKind registration (what kind of
thing this plugin produces), and a typed Source-write API (the index rows it emits).
Fixtures and tests are framework-enforced so quality doesn't degrade as the catalog
grows.
Small on purpose. Most plugin contracts are too big for an agent to fill in correctly without tribal knowledge. inseam's is the opposite — opinionated, minimal, shape-checked. You describe a tool you use, an agent writes the plugin, the test suite tells you when it's right.
The retrieval layer underneath is one ranking algorithm over a normalized corpus — not a fan-out of N source-specific searches glued by an LLM. Cheaper at inference, better recall, and introspectable. See how it works for the index shape.