Your custom agent, supervised by our team.
AITWIRE’s team sets up your custom agent, checks its work and brings the decisions that need you to you — the same oversight as AITWIRE Managed, built on your scope.
How AITWIRE Managed worksAITWIRE Custom
A custom agent, built around how your business works.
An agent configured around your workflows, approved knowledge, integrations, channels and approval requirements — and, where you run AI inside your organisation, that AI tested against the knowledge you approve.
A custom agent can be managed by you or by AITWIRE. Customisation and day-to-day management are separate choices.
A scoped engagement
Agree what a useful answer must do.
- Your approved knowledge
- Choose the policies and sources your agent, and any AI you run, should use.
- Your test questions
- Agree expected behaviour and what a reviewer needs to see.
- Your evidence
- Keep answer records, changes and later checks together.
Designed around your environment
What a custom agent can include.
Each item is agreed in the engagement scope. Nothing on this page is switched on from a form.
1
Your approved knowledge and test questions
The facts, policies and sources your agent — and any AI you run — may work from, and the questions that matter most to your customers. Agreed with you, approved by the people you name, and kept as the record every later check is read against.
2
Your channels and connections
Which channels your agent may use and which systems it may connect to — your website, your business profiles, your advertising account, and the internal systems you choose. Every connection or build is confirmed in the engagement scope; nothing is assumed.
3
Your approval requirements
Who approves what: the owner or manager, a professional approver where the rules require one, a second approver for sensitive wording, and money as its own permission. The workflow follows your rules, not a default.
4
Private AI checks
Where you run AI inside your organisation — copilots, support bots, website chat — it is tested against the information you approve and the claims you prohibit, with the model actually reported and the knowledge version used recorded with every check.
5
Evidence for your reviewers
Answer records, the changes made, and the later checks, kept together in the reporting format your reviewers need. Each measurement carries its date, sample size, the AI system asked and the model it reported.
Two separate choices
Customisation, then day-to-day management.
Your custom agent, run by your own team.
Your people set its goals, review its recommendations and approve its work, with the approval rules and permissions agreed in the scope. How day-to-day management is handed over is part of the engagement.
Managing an agent without an engagement isn’t open yet — the self-service waitlist is the place to be told when it opens.
Join the self-service waitlistBring the problem and the environment.
Tell us which answers matter, which systems are involved and who needs to approve what. We agree the scope — and whether you or we run the agent day to day — before anything is built.
Talk to us