IntegAI Policy Advisor
AI that knows its job, its sources and its limits.
Give one team a Policy Advisor for one defined job. Configure the sources it may use, the questions it may handle, the limits it must respect and the point at which a person must take over.
Start with one controlled use case. Prove it before making a wider commitment.
- A defined job
- Approved sources
- Explicit limits
- Human accountability
- Decision evidence
Live external pilot
Buyer-owned deployment path
Controlled evaluations available
The model is not the job.
Most AI projects begin by choosing a model and asking what it can do. That creates a broad capability before the organisation has defined the work, the evidence or the limit.
IntegAI starts in the opposite direction. Define one job. Approve the sources. Set the boundary. Name the person who remains accountable. Then configure the AI around that operating position.
Policy alone does not control an AI answer.
Policies, training and acceptable-use documents matter. They do not, by themselves, determine what an AI system may answer at the point of use.
A team still needs to know which sources are authoritative, what the system should refuse, when uncertainty requires escalation and what evidence must be retained after the interaction.
If an AI system cannot state its job, its sources and its limits, the organisation is asking people to supply the governance after the answer has already been produced.
One governed product
A Policy Advisor configured around the work.
IntegAI Policy Advisor gives a named team a controlled way to use approved organisational knowledge without turning a general-purpose model into an unbounded authority.
A defined job
The Advisor is configured around a specific organisational task, audience and decision boundary.
Approved sources
The organisation defines the knowledge sources the Advisor may use for the configured use case.
Explicit limits
Requests outside the configured role can be qualified, refused or escalated.
Human accountability
A named person or role remains responsible where judgement, approval or exception handling is required.
Visible basis
Deployed Policy Advisors can show the source sections and configured basis used to form a response.
Decision evidence
Requests, sources, governance outcomes, timestamps and identifiers can be recorded for review as configured.
The chat on this site is a limited public demonstration. Deployed Policy Advisors show source sections, escalation and evidence records as configured.
A controlled supplier-policy use case
The useful answer may be: not yet.
A project team asks whether a new supplier can be approved.
Without a configured Policy Advisor
A general-purpose assistant may summarise common supplier checks, mix general knowledge with organisational policy and leave the user to decide what applies.
With IntegAI Policy Advisor
- ■The Advisor works from the approved supplier policy and evidence requirements for that team.
- ■It identifies the relevant source sections.
- ■It distinguishes evidence received from evidence still missing.
- ■It does not issue the final supplier approval if that authority remains with a named person.
- ■It escalates the exception according to the configured process.
- ■It preserves a record of the request, source basis and governance outcome as configured.
The value is not a more confident answer. The value is a controlled route to the right decision-maker.
Configure the work before you configure the AI.
Define the job
Agree the team, task, intended users and measurable outcome.
Approve the sources
Identify the policies, procedures, standards or records the Advisor may use.
Set the limits
Define prohibited requests, uncertainty thresholds and approval boundaries.
Govern each request
Apply the configured controls before a response is released.
Answer, qualify, refuse or escalate
Produce the permitted outcome and direct exceptions to the accountable person.
Preserve evidence
Record the request, sources, governance outcome and relevant identifiers as configured.
A smaller first decision.
Buyers do not need to approve an organisation-wide AI programme to learn whether governed AI can help.
They can start with one team, one approved document set and one measurable problem. The evaluation either produces enough evidence to justify a deployment decision or it stops without creating a production dependency.
A defined scope for procurement and risk review
A controlled source set
Named acceptance criteria
Human ownership of exceptions
Evidence for an end-of-proof decision
A buyer-owned or embedded path if the use case succeeds
Five questions before AI touches the work.
1.What exact job is the system being asked to do?
2.Which sources is it allowed to use?
3.What must it refuse or escalate?
4.Who remains accountable for the outcome?
5.What evidence must be retained?
If those five answers are unclear, the organisation is not ready to automate that use case.
Start where the evidence already exists.
A strong first use case has a named team, an approved document set, repeatable questions and a clear point at which a person must decide.
It should be useful enough to measure, narrow enough to control and non-critical enough to evaluate without production dependency.
Good starting shapes
- ✓Policy interpretation for one team
- ✓Submission completeness checking
- ✓Approved-procedure guidance
- ✓Evidence-gap identification
- ✓Controlled internal knowledge access
Poor starting shapes
- ✕An unrestricted assistant for the entire organisation
- ✕Automated legal or regulatory decisions
- ✕Production-critical actions during evaluation
- ✕Use cases without an accountable owner
- ✕Use cases that require uncontrolled sensitive data
30-day Private AI Proof
One team. One document set. One measurable problem.
Define the use case, configure the Policy Advisor, exercise it against agreed scenarios and finish with a documented decision: proceed, revise or stop.
Scope, data, access, support and acceptance criteria are agreed before the proof begins.
How IntegAI works
The governed core underneath
One Policy Advisor is the starting point, not the limit.
The IntegAI platform provides the configurable control, approved knowledge access, human approval boundaries, evidence handling and deployment choices underneath each Policy Advisor.
After a successful proof, the next discussion may be buyer-owned deployment, embedded integration, white-label configuration, platform licensing or a strategic relationship.
Independent evaluation, with a clear boundary.
The live IntegAI web and API service was independently exercised by TUS/SRI COMAND across functional, safety, consistency and auditability scenarios under defined test conditions.
The evaluation was not a legal certification, regulatory approval, penetration test, source-code audit, or compliance certificate.
IntegAI Academy
Learn the governance before you configure the system.
Practical training for leaders and teams working with the EU AI Act, NIS2, governed AI and accountable deployment.
Explore the AcademyStart with the use case, not the transformation programme.
Tell us the team, the document set, the problem and the person who remains accountable. We will determine whether it is suitable for a controlled proof.