togan

How it works

One AI relationship, everywhere

You already use several AIs. Each one meets you as a stranger. Togan builds a single, evolving understanding of you that belongs to you, and lets any model read it. Here is the shape of it.


The shape of it

Signals in

Everything you already do with AI

Your conversations, the history you import, the docs you write. Captured as raw signal, on any surface.

it gets distilled, not just stored

The layer that’s yours

An evolving understanding of you

Not a list of stored facts. What you value, how you think, how you work, with the evidence behind every piece. You can see it and correct it. This is the part you own.

at the moment you ask

Read path

Any model gets a compact picture of you

Not a data dump. A small, relevant projection: what matters for this task and how to work with you. The model pulls specific facts and evidence only when it needs them. Understanding first, facts on demand.

↻ It compounds.  The model’s answer and your next turn become new signal, and the understanding gets sharper. Every conversation makes the next one better.
For builders: the full data flow

In practice

01

Bring what you already have.

Import your history from the AIs you use. No starting from a blank profile, and no questionnaire.

02

It learns you, in the open.

Togan builds the understanding and shows its work. You can see what it believes, why, and fix anything that’s wrong. It stays fixed.

03

Use any model, warm.

Switch between models and surfaces and keep one continuous relationship instead of three strangers. The context follows you.


The proof

The first session

Twenty minutes in, it knows you in a way no chatbot does.

Import your history and Togan reflects you back: your patterns, your priorities, how you actually work. Warm start, not cold start. That first-session moment is the whole bet, and it is what we are building the product to deliver.

honest about where it goes: cloud-hosted, zero data retention, yours to export or delete.


Why this person

A twenty-year through-line, not a pivot.

The question underneath Togan is one Mike Holzer has been working since 2005: how do you make information about a person genuinely retrievable, not just stored. A 2007 patent filing on viewing an instant-message archive describes the same retrieval pattern Togan ships in 2026, nineteen years later. Same instinct, different substrate.

Along the way: design and product leadership at Yahoo and Xobni, a decade at Google (including Google Assistant and Payments), and three and a half years leading HubSpot’s AI-first transformation of a 270-person UX organization. Full-time on Togan since 2026.

  • 5 patent filings, 4 granted, 3 active
  • Yahoo · Xobni · Google · HubSpot
  • 20 years on one problem

Come in early

Not the average. Be first.

Togan is being built right now. Leave your email and you’ll be among the first to get in.

Join the waitlist