$ inseam · the problem · what changes · how it works · status · github v0.x · open source · being built in the open
inseam

AI that knows what you know.
Without uploading any of it.

Every AI tool that wants to help with your actual work asks for the same thing: hand over your inbox, your files, your notes, your accounts — into their cloud, on their hardware, priced by the seat. Inseam is the open-source alternative. Your data stays where it lives. You run it. You own it. Built by the people who'll use it.

v0.x · MIT-licensed · built in the open · contributions welcome
01The shape of the problem

Today, you get two bad deals.
Pick your poison.

You want AI to actually help with your work — your real work, the kind that lives across your inbox, your drive, your notes, your tickets, your account with that one vendor your team somehow can't quit. The market currently offers two ways to do that. Neither is great.

Deal #1 — Wire it directly
  • Each tool has its own login, its own permission grant, its own search box, its own quirks.
  • The AI only ever sees what each search box hands back. Usually badly.
  • A question that touches three tools means three half-answers. You stitch.
  • The tool your team actually depends on? Probably not on the supported list. Wait in line.
Deal #2 — Upload it all
  • A second copy of every email, file, message — chopped up and stored on the vendor's hardware.
  • Per-seat pricing that turns into real money the day you finish the trial.
  • Permissions mean "is this person inside our workspace?" — useless for talking to customers, clients, vendors.
  • Whatever the last crawl picked up. Edit a doc this morning, the AI is still answering yesterday's version.

The thing both options assume is that somebody else has to hold your data for you to get any value out of it. They don't. That assumption is what Inseam is for.

02What changes

Things you can do
that aren't possible anywhere else.

Not "we do them better." We do them at all. Here's what becomes a normal Tuesday once your AI talks to your stuff through Inseam instead of through a SaaS that's holding it hostage.

01One AI, all of it
"What did I tell the client about the deadline?"

The answer is somewhere across your email, a Slack DM, the contract draft in Drive, and a sticky in Notion. Today you get four search boxes. With Inseam, you get the answer — pulled from all of them, in one conversation, with the actual text in front of you.

02Your customers can use your AI
"What's the status on my order?" — asked by the customer.

A customer signs in. Your AI answers about their emails, their invoices, their account — and can't see anyone else's, ever. Today this is built bespoke per company, or — far more often — not built at all, and the chatbot stays a useless toy that can't reference anything customer-specific.

03Always the version that exists right now
Not whatever last night's crawl picked up.

Edit a doc at 9:00. Ask the AI at 9:01. You get the 9:01 version. Cloud-RAG systems ingest on a schedule and your "intelligent assistant" is forever a few hours behind. It's one of the things people quietly hate about every existing tool.

04Cancel a service, keep what you knew
Rotate, switch, migrate — without forgetting.

Move from Gmail to Fastmail. Rotate an API key. Drop a SaaS you no longer pay for. The memory of who said what, what was where, when it happened — that stays. Everywhere else, pulling a connector either purges the data or leaves dead weight behind.

05The tool your team actually uses
Even when no vendor will ever support it.

Internal CRM. Niche project tracker. The thing one team built and the rest of the company adopted. Inseam is open, with a plugin model designed so anyone — including an AI helping you — can add support in an afternoon. Your stack doesn't have to wait on someone else's roadmap.

06Runs where you tell it to
Your laptop. Your server. Both. Whatever.

Quietly on a laptop for a personal setup. Inside a team's own infrastructure for a small company. On a self-hosted server you already pay for. There's no vendor cloud you have to route through; there's no "enterprise edition" gating the feature you want. It's open source. You run it where it makes sense.

Curious how it actually works under the hood? The architecture page walks through it.

03Straight talk

What you're getting.
What you're not.

We've been around long enough to know that the marketing page is usually where software lies to you. So instead of adjectives, here's the actual deal.

It is
  • Open source. The kind you can read, run, and fork.
  • Yours to run — your machine, your server, your call.
  • Built so the community extends it, not so we gate it.
  • Real about who-sees-what. Per person, not per workspace.
  • The layer between AI and your stuff. Not another place to put it.
  • Something that gets out of the way once it's working.
It isn't
  • A cloud we host. A SaaS we bill you for.
  • A second copy of your documents on our hardware.
  • A chatbot. We don't ship the AI; we ship what the AI talks to.
  • A list of supported tools you have to pray yours is on.
  • An enterprise upsell gating the features that matter.
  • A pivot waiting to happen — there's no investor to please.
04Where we are

Early. Honest about it.

This isn't a "coming soon" page from a stealth startup. It's a project being built in public, on nights and weekends and afternoons, by people who want this thing to exist for themselves. Some of it works today. Some of it is being designed right now. Some of it is on the list.

If you'd want to use this — or contribute to it, or push back on a design while it's still on paper — now is the time to show up. The work happens in the open under /architecture. Issues and pull requests are read. Cheerleading isn't necessary; honest argument is welcome.

seam.status utc · last sync now
$ inseam status --verbose
[x] logging in to your accounts (the auth layer) landed
[x] storing things locally (on a laptop) landed
[x] storing things on a server you control landed
[x] the visual language you are reading right now landed
[~] what an AI can actually ask for in design
[~] talking to it over the network in design
[~] a web interface for humans in design
[ ] permissions — per-person, not per-workspace queued
[ ] first plugins — gmail and google drive queued
[ ] first plugin — files on your own computer queued
[ ] a guide for writing your own plugin queued

The community is small and the tent is open. If a tool you depend on isn't on the list, that's where your plugin ends up. If a design decision looks wrong, the design docs are still drafts — say so. The repo is the front door.