Otto is a self-hosted AI platform: it ingests your documents and data, stores them across four purpose-built databases, remembers under governance, and routes work across AI models you control. Everything below is the running system — not slides.
| Min | Topic | What you'll see |
|---|---|---|
| 0–6 | The idea, and the architecture | Why a platform rather than a chat window, and the five layers that make it up — one diagram, no jargon. |
| 6–8 | How knowledge gets in | Ten intake channels — upload, scanner, email, RSS, podcasts, notes — converging on one enforced pipeline. We'll look at a real run of it. |
| 8–16 | A document's whole life (live) | The core of the session. We load a company handbook and ask it questions; Otto answers with citations you can open. We pin the document for follow-ups. Then we delete it — and watch Otto correctly say it no longer knows, while the original conversation still remembers. Retrieval, sourcing, and forgetting, end to end. |
| 16–23 | How it learns and how it chooses | Memory you can decay, correct, and switch off. Per-person, per-domain voice that evolves nightly — with a human approving every change. And model routing: cheap models for cheap work, frontier models where they earn it, swappable in one config file. |
| 23–29 | Running it for real | Household modules built on the same foundation (meals, chores, tasks), a cost dashboard showing exactly where every fraction of a cent went, and an honest look at what it would take to bring this to enterprise scale. |
| 29–30 | Questions — and a login | You leave with guest access to a live instance, so you can try it yourself. |
Bring questions. The section on limitations is deliberate — we would rather you hear the gaps from us, with the work each one would require, than discover them later.