Otto
Smart routing · real memory · checked answers

The assistant that picks the right model for every task.

Otto reads what you're asking, routes it to the open-source model best suited for the job, remembers what matters to you, and checks its facts against the live web.

Open Otto
Otto is in pre-alpha — the core works today, with a lot more still to come.

Right model, every prompt

Code, research, vision, or a quick reply — Otto classifies each request and routes it to a model that fits, per prompt, not per chat.

Memory that stays

Otto owns your memory — facts and preferences persist across conversations, ready before the model even starts thinking.

Answers you can trust

The Truth Gate lets you dial how hard Otto fact-checks — from fast and loose to web-verified with citations.

What Otto can do

What works today — and what's next

Smart routing

Specialised open-source models for code, reasoning, summarising, vision and strict JSON — chosen automatically.

Web research

Searches and reads live pages, then folds the evidence into its answer with sources.

Persistent memory

Remembers your context across every chat. Hotfixes landing soon will clean up how those memories are stored and surfaced, so recall keeps getting sharper.

Truth Gate control

Set a fact-checking level per chat or account — 0 for speed, 10 for verified.

Integrations coming soon

Google Calendar and Gmail support is on the way, so Otto can work with what's already on your plate.

Graceful fallback

If a model stumbles, Otto quietly retries with a compatible one — same answer shape, no drama.

Feedback built right in

Since Otto is early, telling us what's wrong is a first-class feature: feedback, bug reports, and feature requests are all built into the app. Say what broke or what you want next, and it goes straight to the people building Otto.

Ready when you are.

Open Otto and start with a single question.

Open Otto
Otto · LLM gateway — routing · memory · web evidence · truth gate