Use case · KYC onboarding
Agent Serve reads each new KYC pack, checks it against your onboarding policy and lists what is missing, so analysts spend time deciding.
A KYC onboarding review use case for Agent Serve, the AI workspace where teams build, deploy, and manage AI agents: a workflow reads identity and address documents, checks them against the onboarding policy in a knowledge base, records the result in a table and waits for a KYC analyst to decide.
At a glance
The problem it removes
A new business customer emails a passport, a utility bill and a company extract. An analyst copies the details into the case system, checks each document against the policy by hand, and asks for what is missing days later. Queues grow, and file reviews find the same gaps again.
How it runs
What arrives, what Agent Serve does with it, and where a KYC analyst makes the call.
A new application lands as a row in the onboarding table, with the passport, utility bill and company extract attached.
AWS Textract reads the identity document, and an agent pulls the name, date of birth and address out of each file.
The agent compares the pack with the onboarding policy in the knowledge base and lists each missing or out-of-date item, with its clause.
The run pauses at a Human step. The analyst reads the note and the documents, then records a decision and a comment in the resume form.
Before and after
| Work | Before | With Agent Serve |
|---|---|---|
| Reading the pack | BeforeAn analyst retypes names, dates and addresses from each PDF. | With Agent ServeAgent Serve reads the documents and fills the case record. |
| Checking the policy | BeforeThe analyst checks a printed checklist from memory. | With Agent ServeEach gap is listed with the policy clause it comes from. |
| Chasing what is missing | BeforeThe request goes out days later, after the first look. | With Agent ServeThe missing items are known as soon as the pack arrives. |
| Knowing where files are | BeforeStatus lives in a spreadsheet someone updates on Fridays. | With Agent ServeA table and a dashboard show every file and its stage. |
What does the work
An onboarding table holds one row per application, and a new row starts the review.
A workflow reads the documents, checks them and stops for the analyst.
The onboarding policy and procedures sit in a knowledge base the agent cites.
Rules mark higher-risk files, such as complex ownership, for a senior analyst.
A dashboard shows open files by stage, analyst and age.
In depth
Agent Serve puts the documents, the policy answer and the case record where the analyst already works.
Agent Serve keeps each KYC pack in the Library next to the onboarding policy, and a workflow writes its review note beside the documents it read.
An analyst asks Agent Serve Chat what counts as proof of address, and the answer links to the passage in the onboarding policy.
Each step of the Agent Serve review is a block that says what it does, so a compliance lead can follow the logic and change it before it is deployed.
An Agent Serve score adds points for risk factors such as a high-risk country or layered ownership, and the queue filters by the result.
Examples
Three onboarding queues financial services teams run on Agent Serve, each ending with a person’s decision.
An Agent Serve workflow reads the company extract and director IDs, lists missing items with the policy clause, and stops for a KYC analyst.
An Agent Serve workflow checks each merchant’s documents against the onboarding policy and routes higher-risk scores to a senior reviewer.
An Agent Serve schedule picks clients due for review, gathers their records and drafts the refresh note for an analyst to sign off.
Examples show what teams can build in Agent Serve. They describe patterns, not named customer deployments.
Governance
Agent Serve keeps a person on every onboarding decision and limits who sees customer documents.
Security and governanceAgent Serve restricted columns hide fields such as passport numbers from everyone you do not name, in the table and in exports.
The Agent Serve Human block pauses each review until the analyst responds, and the response is kept in the run’s log.
A Guardrails block detects personal data and blocks or masks it before text reaches the next step.
Proposals, approvals, declines and executed actions are written to the workspace audit log, which admins can export.
Admin, write and read roles per workspace, plus permission groups that limit which models, blocks and integrations a team can use.
Sign in through your identity provider with SAML 2.0 or OIDC, and provision people with SCIM.
Run Agent Serve in your own cloud or data centre, next to the systems it reads.
Zero data retention with supported model providers, backed by your own provider keys and agreements or self-hosted models. Some provider features keep data by design, such as Gemini Deep Research, so they are outside zero data retention.
Further reading
The deep dive models what a periodic review backlog costs and what changes, with every figure labelled.
Read the blogPeriodic reviews pile up because every file starts from scratch, even when nothing has changed. Here is a setup that refreshes the data, drafts the file and routes it by risk, so analysts spend their time on the reviews that need judgement.
Retail and business banking · 12 min readOnboarding files arrive incomplete, and reviewers spend their time finding out what is missing. This playbook builds one workflow that checks the file first and routes it by risk to the right person.
Retail and business banking · 5 min readOperations staff ask the same policy questions every day, and a wrong answer becomes a breach. This guide compares four ways to answer them, and shows when a cited, logged Q&A workflow is the right choice.
Banking operations · 6 min readQuestions
No. In this Agent Serve workflow the Human block pauses every review until a KYC analyst responds, and the analyst’s decision is recorded with the run.
An Agent Serve workflow can use the AWS Textract block, which reads text, tables, forms and identity documents, and an Agent block that pulls out the fields the policy needs.
Upload it to an Agent Serve knowledge base or sync it from a source such as SharePoint, Confluence or Google Drive, hourly, daily, weekly or on demand.
Agent Serve applies team space roles to the Library, and row limits and restricted columns to tables, so each person sees only what their role allows.
A team can start with the Agent Serve review workflow, the policy knowledge base and the onboarding table, then add risk rules and a queue dashboard once the first part is trusted.
Last updated
Enterprise access
Agent Serve is available to teams by invitation. Tell us what you want to build and we will set up your workspace with you.