Customer Onboarding Agent · Use case
How can a bank use an AI agent to prepare account-opening files on its own infrastructure?
A customer onboarding agent reads the documents an applicant sends, extracts the details, checks them against the application and your screening service, and prepares the account-opening file. A member of staff approves or declines every account: the agent opens nothing. It runs on your own infrastructure, so identity documents are not sent to a public AI service.
Forms, documents, checks.
All typed in by hand.
- ✗ 01Onboarding staff retype details from identity documents and proofs of address, then check each one against the application and the watchlists.
- ✗ 02Identity documents are personal data the bank must keep inside its own systems, which rules out a hosted assistant.
Where it stops for a person
Always, before an account is opened.Always. A person approves or declines every account, and any watchlist result goes to compliance. The agent does not open an account or reject an applicant.
It prepares the file.
A person approves the account.
Step · 01
The applicant's documents arrive through File Upload.
Step · 02
The agent reads each document and extracts the details.
Step · 03
It compares them with the application and notes every mismatch.
Step · 04
It asks your screening service for watchlist results and adds them to the file.
Step · 05
A member of staff reviews the file and approves or declines the account.
Reads from · connectors
File Upload
Your own APIs
Documents arrive through File Upload and are read with Markitdown. Screening results come from your own screening service, which the agent connects to through your own APIs.
Acts through · MCP servers
Markitdown
Nothing. It prepares the file and lists what is missing or does not match.
What you set up
- List the documents you accept and the fields you need from each.
- Connect your screening service.
- Write test files with known results, including ones that should be flagged.
Where it runs
The agent, its knowledge store and the model run on infrastructure you own or control. You choose the model and serve it yourself, so questions, documents and answers are not sent to a public AI service.
How you get it today
This page is labelled Use case: the agent is built in Katonic Studio. You describe the job, connect what is named above, add test questions and release the agent once it passes them. When the Catalog holds a tested template for this job, this page will carry the Template label and show how to take it.
Does the agent open the account?
No. It prepares the file. A member of staff approves or declines.
Does it do the watchlist screening itself?
No. It asks the screening service you already use and puts the result in the file for compliance.
Where are identity documents processed?
On your own infrastructure, with a model you serve. They are not sent to a public AI service.