In short
Chakaap records each AI-assisted action as a structured entry: what arrived and from whom, what the agent extracted and from which document, which configured rules ran and what they found, what action was taken or proposed, and which person approved it and when. Compliance and audit teams can review that trail per case, per rule or per person, and use it as evidence of how work was done.
Why AI needs a record
Teams already use AI tools to summarise documents, draft replies and check data. The work gets faster, but the evidence disappears: a chat window does not tell an auditor which document a figure came from, which policy was applied, or who decided to act on the result.
For regulated institutions that is a problem. Supervisors increasingly expect firms to explain how automated or AI-assisted processes reach their outputs, and to show that people stay accountable for decisions.
What goes into each record
| Field | Example |
|---|---|
| Input | Email from a branch, 3 attachments, received 14:05 |
| Extraction | Customer name, ID number, transaction amount, with the source file and page |
| Rules applied | Sanctions-list check requested; amount above reporting threshold |
| Action | Case flagged and routed to the compliance officer; draft report prepared |
| Approval | Approved by the compliance officer at 15:20, with a comment |
Because the record is structured rather than free text, it can be filtered and exported: all cases where a rule fired this month, every action a given person approved, or the full history of one customer file.
A worked example
A branch forwards a customer due-diligence file for a new corporate account: a registration certificate, a list of directors and a bank reference letter. Your configured rules require identity documents for every director.
- Chakaap extracts the company name, registration number and the four directors listed.
- It finds identity documents for three of the four directors and flags the missing one.
- It drafts a request to the branch for the missing document and routes the file to the compliance officer with the gap highlighted.
- The compliance officer approves the request. The record shows the input, the extraction, the rule that fired, the draft and the approval, with times.
When the file is reviewed months later, the reviewer sees exactly what was received, what was missing, and who decided what.
Who it is for
- Compliance officers who need evidence that AI-assisted checks were run and reviewed.
- Internal audit teams who sample cases and need to rebuild what happened in each one.
- Risk and governance leads who set which actions may run automatically and which need approval.
- Operations directors who want faster processing without losing accountability.
The record is useful beyond audits: it shows where cases wait longest, which rules fire most often, and where documents are usually missing, so processes can be fixed at the source.
Governance by design
- Controlled actions: you decide which actions run automatically and which need approval.
- Human oversight: decisions with consequences wait for a named person.
- Traceability: every extracted value keeps a link to its source document.
- Operational history: the trail is kept per case, so a file can be reconstructed end to end.
Questions
Does Chakaap make compliance decisions?
No. It prepares the work (extraction, checks against your configured rules, drafts and routing) and records it. Decisions are approved by the people you assign.
What exactly is recorded?
The input received, the fields extracted and their source, the rules that ran and what they found, the action taken or proposed, and the approval with person and time.
Can we export the audit trail?
The trail is structured so it can be filtered by case, rule, person or period for review and reporting.
Does it replace our compliance system?
No. It works alongside your existing systems and channels, adding a record of the AI-assisted part of the work.
Is Chakaap available now?
Chakaap is in early access: it is being rolled out to a small group of institutions, features arrive over the coming weeks, and we set up every early-access account with you personally.
How do we start?
Provision a Chakaap identity on the home page. Solve (free) covers triage, extraction and routing; Sync (NPR 3,999 + 13% VAT per 30 days, paid by Fonepay QR) adds AI tool connections and drafting.
Give your team an operational AI layer.
Start with Solve (free) for document triage, extraction and routing, or Sync (NPR 3,999 + 13% VAT per 30 days, paid by Fonepay QR). Chakaap is in early access.
Create your account →