02 from the closed channel, March 2026

The AI intake gate: before an AI tool reaches company data

Firebase security rules on terminal beside exposed AI chat database records

An AI app is a standing data pipeline to a third-party backend. Teams adopt them to move faster, and an app-store listing passes for a security review. The gate goes up before the first prompt, not after the first breach.

  1. 1

    Enumerate before you govern

    You cannot scope a breach you cannot list. Find every AI tool in use: the expense reports, the browser extensions, the SSO logs, the DNS logs. Each one is a data pipeline to somebody else's backend.

  2. 2

    Block three things on sight

    A tool whose vendor gives no account of its backend. A backend service whose access rules nobody can show you. Adoption nobody tracks, team by team, in the name of moving faster.

  3. 3

    Ask for the rules, not a questionnaire

    Before a tool is approved, get proof of authentication, encryption at rest and logging, and where the vendor runs on a backend service such as Firebase, the access rules themselves. A Realtime Database rule that reads ".read": true is a public database, whatever the questionnaire says.

  4. 4

    Register and connect

    Every approved tool goes into one registry, behind SSO, with its logs going to the same place as everything else. A tool that cannot use SSO or central logging is blocked at the edge.

  5. 5

    Check again

    Vendors change their backends without telling you. Re-run the check on a schedule, and when a tool fails it, quarantine it until it passes.

Two rules for procurement

  • A vendor that cannot show how its backend is built and secured does not get a purchase order.
  • A tool that cannot use SSO and central logging is blocked at the network edge or by device management, whoever asked for it.

The liability “Business units choose their own AI productivity tools.”

The control “Endpoints the architecture board has not cleared are blocked on the network.”

The order to do it in

  1. Publish the list of sanctioned AI tools, and block the rest at the network and on managed devices.
  2. Check the backend of every sanctioned tool, its database rules included, and repeat the check on a schedule.
  3. Put the gate into procurement and give it an owner by name.

The board will not ask what the vendor did wrong. It will ask why your network allowed the connection.

Reports differ on whether reading the data needed any authentication. The long version is the post The Intake Gate Your CISO Is Missing.


Vladimir Mikhalev

Docker Captain  ·  CNCF Ambassador  ·  IBM Champion

The Verdict — production-tested analysis on YouTube.

The Verdict

Inconvenient truths about shipping in the AI era

Container security, platform engineering, and the agentic shift — tested in production, argued without the hype. The verdict reaches your inbox the moment there's one worth sending.

The other blueprints

The AI intake gate: before an AI tool reaches company data
https://heyvaldemar.com/blueprints/ai-intake-gate/
Author
Vladimir Mikhalev
Published
2026-11-13
License
CC BY-NC-SA 4.0