Skip to content
What your AI is fed

Analect Governance

Analect Governance puts file-level control over everything the other workflows store. Each document gets a sensitivity label worked out from what it contains, such as how many people it names and whether it carries account numbers, and that label sets who may open it, what is withheld from whom, and what is logged.

On the platform, with access set up by ETT Last reviewed by George McKenna, Chairman & CTO, ETT.

The question it answers

Who may open each document, how sensitive is it, and who has looked?

What goes in

  • The documents your account has stored through the other workflows.
  • Your rules: which kinds of detail to withhold, from whom, and on which topics.

What comes out

  • A sensitivity label for every document: public, internal, confidential or restricted, with the counts it rests on shown.
  • Sharing by role: read and relabel, read only, read with sensitive details masked, or see that it exists and nothing more.
  • Redaction by kind of detail and by topic.
  • Share links that always expire.
  • A log of who opened what, and when.
  • An emergency route for urgent access that is recorded, and that never opens restricted material.
Built to be checked

How is it measured?

  • The label is a count, not a guess. How many people a document names and whether it carries account numbers are counted from the document itself and shown to whoever accepts the label.
  • A label is not picked off a list at save time, where everything ends up as internal. A person with the right role can still change it.
The limits

What does Analect Governance not do?

  • It does not yet stand in front of a model call. Today it governs who may read stored material.
  • It does not label material held in your other systems. It works on what Analect stores.
  • It does not treat a label as a legal classification. It is a working label your people can see the reasons for and change.
Where it sits

Where does Analect Governance fit in an AI pipeline?

Beside it, not inside it. Analect does not sit in the path of your AI requests, and you do not change your models or tools to use it. Analect Governance works at the point marked below.

  1. 01

    Before you build

    What the work should cost, and on which model.

  2. 02

    Prepare the documents

    What is in them, what they cost to hold, who may see them.

    Analect Governance works here.

  3. 03

    Choose what the model is given

    Whether sending less still gets the same answers.

  4. 04

    Run

    Your own models and tools. Nothing of Analect's sits here.

  5. 05

    Check the output

    Whether each claim is supported by its source.

  6. 06

    Prove it over time

    Evidence an auditor can follow, for one system over a period.

How to get it

How do we start using Analect Governance?

Analect Governance is on the platform. ETT sets up access on your own material. Accounts are set up by ETT, and each account sees only the workflows its work needs.

Also under what your AI is fed: Analect Explorer, Analect Store, Analect Proof. All workflows

Questions

About Analect Governance

How does Analect Governance decide a document's sensitivity?

From what the document contains. It counts things such as the people a document names and the account numbers it carries, derives one of four labels from those counts, and shows the counts to whoever accepts or changes the label.

Does Analect Governance control what is sent to an AI model?

Not yet. Today it governs who may read stored material: who may open a document, what is masked for whom, and what is logged. It does not stand between a document and a model.

Can a document be shared with somebody without showing them everything?

Yes. A share can give the whole document, the document with its sensitive details masked, or only the fact that it exists and how it is labelled. Share links always expire.

See it on worked material, then on yours

Thirty minutes with the team that runs Analect: the AI Spend Forecast and the Accuracy Evidence File on worked examples, and what either would look like on your bills or your outputs.

Book a demo