Governed research jobs

Deep research in a box your CIO and CCO can sign for.

Start in Slack or the AI chat your team already uses. You keep curiosity and review. The box keeps the data and the audit trail. First deliverable in under an hour.

It runs against approved data: Bloomberg, FactSet, your files and git. You review and publish.

For

Investment teams at funds with 50+ people, roughly $2B+ in assets, and an IT team of one to ten.

Not for

Not a chatbot. Not a standalone gateway.

How a job runs

Four steps. Then the work stays put.

  1. 1

    Start where you already work

    Name the company or question in Slack or the AI chat you already use.

  2. 2

    The job runs in a locked-down workspace

    devBox plus Harness. No coding workspace required of the analyst who started it.

  3. 3

    Only approved models and sources

    Bloomberg, FactSet, internal files, git. Identity and allow-list on every request, via mGateway.

  4. 4

    A human reviews and publishes

    First signed artifact in under one hour, started by a non-engineer.

The workspace keeps the work and ranks what to do next, inside the same controls. Full path.

The environment

One path. Not a menu of products.

Start in Slack or the AI chat the desk already uses. The job runs in a locked-down workspace against approved data. Policy sits on every model and tool call. You review and publish.

Who it is for

  • Head of Tech who cannot staff an internal agent platform
  • CIO who needs more coverage without leaking the book into a public model
  • Analysts who should start work without owning the toolchain

Who it is not for

  • Teams that only want a better chatbot
  • Teams that only want an AI gateway
  • A firm-wide unscoped "second brain"

Need a map of shadow AI traffic first? That sits in the same control plane. We do not sell it as a separate product.

See a one-hour research run.

Book a working session. An analyst starts the job. You review the first deliverable.