Datacat

Product catalogue / Self-hosted AI delivery agents pilot

Pilot

Self-hosted AI delivery agents pilot

Agents that pick up work items, implement in a pooled sandbox, run tests, and open a merge request — with policy gates and evidence your AI governance forum can review.

Early access — pilot places available. Models and runners stay inside your boundary; Datacat configures agents, policies, and integration with your existing OmniStudio DevOps platform where present.

Fixed-scope delivery:Self-hosted AI delivery agents

Inside your boundary

Ticket intake from your work tracker, execution on infrastructure you control, and merge requests your teams already review — with logs retained for audit.

  • GitHub, GitLab or Azure DevOps native flows
  • Human approval before promotion beyond sandbox
  • Evidence pack templates for internal AI risk reviews

Built for OmniStudio change

Agents follow the same deploy and test paths your engineers use — including dual-runtime steps during migration — rather than a generic code assistant bolted on the side.

Frequently asked questions

Where does the model run?
On endpoints you designate — for example a private cloud model endpoint. Datacat does not require sending org metadata to a multi-tenant vendor AI service for the pilot.
Does this replace developers?
No. It accelerates well-scoped tickets with human review on every merge request. Policy gates block promotion when tests fail or risk rules trip.
Which ticket systems are supported?
Pilots can start with Jira, GitLab issues, or Azure DevOps work items. Tell us your stack on the contact form and we will confirm fit.
Is this IRAP certified?
We do not claim formal assessment labels. The design target is auditable, self-hosted operation so your security team can evaluate against your own controls.
Can we combine this with the DevOps platform offer?
Yes — most regulated buyers want sandbox pools and release gates in place before agents touch production-bound paths.

Book a fixed-price assessment of pipelines, pools, and AI governance before the self-hosted agent pilot.

Book a fixed-price assessment