Planned

AI Agents

Task-specific agents that prepare work, never authorise it.

Status shown is the honest current state of this area: Planned. Nothing on this page implies a certification, licence, partnership, government approval or endorsement.

What this service is

AI Agents are narrowly scoped assistants built for a defined task — drafting, checking, summarising or preparing a case — operating on reviewed sources and inside the same permission model as the rest of the platform. An agent is never the authorisation authority.

Core capabilities

Task-scoped agent design with explicit boundaries
Grounding on reviewed sources with citation
Permission-aware execution inside existing access control
Human approval and takeover at decision points
Prompt-injection boundaries and logged actions

Typical use cases

Drafting a first response for a support agent to review
Summarising a long case file for a human decision-maker
Preparing a checklist from documents before a person signs off

Benefits & outcomes

Faster preparation with the decision still made by a person
Consistent output format across a team
A logged record of what the agent did and on what basis

Delivery & process

Step 1
Requirement review
A written review of the requirement, data involved and who needs access.
Step 2
Scope & proposal
A scoped proposal issued as a referenced BD AI document that can be verified publicly.
Step 3
Build & review
Implementation under BD AI standards for provenance, access control and audit logging.
Step 4
Handover
Documentation, access handover and an agreed support arrangement.