Use cases/Financial services/KYC and onboarding
AI KYC onboarding for Australian fintechs and banks
KYC and onboarding means each new customer or merchant reaches the analyst as one file. Identity documents, company records and beneficial owner details are read into it, the ABN and ASIC register checks are recorded, and the missing items are listed. For a merchant, the application is also checked against the merchant website and your risk policy, and a draft risk rating is attached. The onboarding analyst approves every customer.
Talk to our team →- Supports
- Onboarding analyst
- Reads
- ID documents, Company extracts, Beneficial owner forms, Merchant websites, Risk policy
- Runs on
- Your servers, on premises
- Connects to
- FrankieOne, GBG greenID, Sumsub, Equifax IDMatrix
- Decides
- A person, every time
Architecture
Everything runs on your servers
We build it on an open-weight model such as Llama or Mistral, served with vLLM on NVIDIA GPUs inside your own environment. Documents are read into PostgreSQL, your risk policy is indexed for retrieval, and every field in the file links to the page or register entry it came from. The log records what the system read and drafted.
- SourcesID documents / Company extracts / Beneficial owner forms / Merchant websites / Risk policy
- Your serversRead the documents → Check the registers → Draft the risk rating
- ReviewOnboarding analyst
- OutputOne file per customer / Register checks recorded / Merchant risk rating draft / Missing items list
It takes four jobs off the onboarding analyst
- 01
One file per customer
Passports, licences, company extracts, trust deeds and beneficial owner forms are read into one structured file. Each field links back to the page it came from.
- 02
Register checks recorded
The ABN and the company details on the ASIC register are checked against what the applicant declared. Each difference is listed with the register entry beside it.
- 03
Merchant risk rating draft
The merchant application is read against the merchant website and your risk policy. The draft rating shows the policy clause and the page behind each factor, and the underwriter accepts or changes it.
- 04
Missing items list
The documents and fields each customer type needs are compared with what arrived. The gap goes to the analyst and into the request to the applicant.
It goes live in five steps
- Step 01
Collect past onboarding files
Approved and declined files with the documents received, the checks run and the risk rating given. This is the test set, and access to it is restricted to the people who already have it.
- Step 02
Write down the onboarding checklist
We sit with the onboarding analysts and list the documents each customer type needs and how a merchant is rated. The checklist and the risk policy are written down and encoded.
- Step 03
Test against past files
The system builds files and draft ratings for applications that already have an outcome. Analysts compare each file with the one they built by hand and mark what was missing or wrong.
- Step 04
Run in shadow
Analysts onboard customers as usual while the system builds files beside them. Nothing reaches the customer record until an analyst accepts it.
- Step 05
Analyst approves and the log records it
The analyst approves, declines or sends back each customer. The underwriter accepts or changes each merchant rating, and the log records what the system read and drafted.
Software or a custom build
Identity verification and screening come from products such as FrankieOne, GBG greenID, Sumsub and Equifax IDMatrix, and a custom build does not replace them. Use their onboarding flows first if your customers are mostly individuals with standard identity documents. A custom build earns its cost when files mix trust deeds, company extracts and merchant websites, when the risk policy sits in documents that analysts apply by hand, or when onboarding files cannot leave your environment.
FrankieOne / GBG greenID / Sumsub / Equifax IDMatrix / LSEG World-Check / Salesforce
In practice
We build this on servers you own. The model, the document index and the logs sit inside your environment, and identity documents are not sent to a public AI service. The retrieval pattern is covered in RAG on private infrastructure, and the build falls under AI agents and workflow automation.
The AML/CTF reforms changed AML/CTF program and due diligence requirements for current reporting entities from 31 March 2026, unless the transitional rules defer them. The AUSTRAC page on the reforms sets out the start dates. The AUSTRAC page on the transitional rules sets out what is deferred, such as initial customer due diligence until 31 March 2029. The system applies the checklist your AML/CTF program sets for each customer class, and your compliance officer decides how the program applies.
The alert side of the same work is covered in AML alert triage. Where the human step sits in a build like this is covered in where to put the human. The wider picture for payments and fintech and banking has its own pages.
Questions
What is KYC onboarding automation?
It is the reading and checking that happens before an analyst approves a customer. The system reads the documents into one file, records the register checks and lists what is missing. For a merchant it also drafts a risk rating against your policy. The analyst approves the customer.
Does the AI approve customers or set the final risk rating?
No. The onboarding analyst approves every customer, and the underwriter accepts or changes each draft merchant rating. The system shows the policy clause and the source page behind every part of the draft.
Does this replace our identity verification product?
No. Products such as FrankieOne, GBG greenID, Sumsub and Equifax IDMatrix keep running the identity checks they run today. What we build reads the rest of the file, assembles it in one place and lists what is missing.
How does it fit the AML/CTF reforms?
We build to the customer due diligence steps your AML/CTF program already sets out. The system gathers records and lists gaps against that program. Your compliance officer decides how the program applies to each customer class under the reforms and the transitional rules.
Industries where this comes up
Payments and fintech / Banking
More financial services use cases
Document search / Credit file assembly / AML alert triage / Claims triage / Month-end close / Audit evidence matching / Advice file drafting / Lease abstraction / Complaints handling / Regulatory reporting / Underwriting submissions / Board and management reporting
Before you call
Tell us where your hours go.
We map the work, size the hardware and price what is worth automating.
Talk to our team →