Togno Labs · AI systems in production

Move from AI to governed AI.

Adopting AI is the easy half. The hard half is being able to account for it: what is in use, who decided, and who answers when someone asks. We build the systems that make that answerable, and they run on infrastructure you already own.

Method

Your infrastructure. Your keys. Your signature.

What we build is installed where you already operate, in accounts you control. Your data does not move to us in order for us to work on it, and there is no console of ours that you have to log into afterwards.

The agent is how the work stays affordable. The signature is what you are buying.

You close the account, the agent stops, and everything keeps running on your server, because it was always yours. No migration, no export, no negotiation.

How a release moves
  1. Prepared

    The change is prepared as a reviewable diff, on your infrastructure.

  2. Checked

    It is checked automatically against a published conformance spec.

  3. Signed

    A named person reviews it and signs. Nothing moves without this step.

  4. Published

    Only then does it reach a parent, a board or a public page. The approval record stays with you.

One of the tools

Built, shipped, and reading real data.

Below is output from the AI exposure scan, a free tool inside AIP. It matches network activity against a maintained signature set and flags AI features that arrived inside licences an institution already pays for. The set is versioned and dated, and this page generates the figures from it when it is built.

Example terminal output: the AI exposure scan loading its signature set, listing how many services it covers by risk band, and naming three AI features embedded in products schools already licence.
Example output

This is a sample run, not a live process on this page.

$ aip exposure-scan --signatures 2026.09.2   signature set loaded .... covers 64 services    19 rated high    33 medium    12 low   AI features embedded in products already licensed: 10    Google Workspace for Education -> Gemini      Often enabled by default on some editions. Verify rather than assume.    Microsoft 365 -> Copilot      Arrives by tenant update.    Canva for Education -> Magic Studio      Arrived by product update, not purchase.   set 2026.09.2, updated 2026-09-19  --  A domain contact is EVIDENCE OF A CANDIDATE, NOT PROOF OF USE.  A single request may be an advert, an embedded widget, a link preview, or one curious click.  Every match is a review queue item for a human to confirm.

The scan runs in the browser against a DNS export or a list of OAuth grants. Nothing is uploaded. It is the same shape as the rest of the work: measure first, on the client’s own data, and let a person decide what it means.

Run it yourself Runs in your browser

What runs

The governance product, and the rest of the shop.

AIP

Institutional AI governance

An assessment, a policy and an operating cadence for institutions that have to account for how AI is used. A free self-check and tool kit to measure where you stand, and paid tiers when you want the work done and signed. The resident agent runs on the institution’s own server and never publishes without a signature.

AIP has its own site, written for the people who have to answer the board.

Go to AIP
  • 20question self-check
  • 64service signatures
  • 9instruments · 205 items
  • 16deliverable samples

How the lab is engaged

Four lines of work.

Education is where the deepest work is, but the method does not change outside it: measure before advising, build on infrastructure the client owns, and leave a named person accountable for anything the system produces.

  • /AI consulting
  • /Software development
  • /Process automation
  • /Custom digital solutions

Constraints

What we refuse to do.

  • Publish a figure we cannot source.
  • Let an agent publish. A named person signs every release.
  • Move your data onto our infrastructure in order to work on it.
  • Report a subgroup of fewer than five people, including when you ask us to.
  • Quote a price we would not honour.

Each one is a constraint we chose, and each one costs us work we could otherwise bill.

Company

Tell us what has to run, and where.

A first conversation is enough to say whether this is work we should take. If it is not, we will say so and point you somewhere better.

Entity
Togno Labs S.A., Panama
Email
info@togno.co
WhatsApp
+57 333 286 1449
Hours
Monday to Friday, 09:00–18:00 UTC
Reply
Replies within two business days