How does a language model system get through model risk management?

It is treated like any other model, with a documented design, an evaluation set from your own cases and accuracy monitored before and after launch. We plan for your validation team to review the design before launch rather than after.

Will the system make decisions on customers, alerts or credit?

Decisions on customers, alerts, sanctions hits and credit stay with your analysts, investigators and credit officers. The systems gather, read and draft, and every recommendation carries its evidence.

Do we have to replace our core, case management or loan systems?

The systems work with the core, case management and loan systems you already run, and the evaluation confirms how before anything is built.

How does this fit our third-party risk management?

We go through your vendor due diligence like any other provider. Because the systems run in your environment and you own the code outright, no vendor holds customer data once the work is finished, which usually simplifies the review.

AI for banks, built for the review work behind every account, alert and loan.

Bluegrass builds production AI systems for banks that take over the review work in financial crime compliance, third-party risk and lending. Each system runs in your own cloud environment and belongs to you.

We have worked with some of the largest banks, and what usually decides whether a pilot reaches production is evidence rather than technology. Every system we build produces the testing, documentation and decision record that examiners and validation teams expect.

Talk to an engineer How the evaluation works

Where a bank's time goes

Financial crime compliance

KYC refreshes, AML alerts and sanctions hits each need an analyst to gather evidence, decide and write up the decision, and every queue grows with the customer base.

Third-party risk

Every vendor and fintech partner sends SOC reports, policies and contracts that someone has to read against your standards before approval, and again at each periodic review.

Fraud and disputes

Fraud cases and card and payment disputes run against deadlines, and each one needs the claim, the transactions and the decision documented.

Lending

Commercial analysts spread financial statements by hand before writing the credit memo, and consumer loans bring files to check before funding and collections, disputes and claims to work afterward.

What we build for banks

Financial crime and compliance

KYC, KYB and EDD reviews

Onboarding exceptions, business customers, enhanced due diligence and periodic refreshes prepared from the documents, with every discrepancy set out for an analyst.

AML alert investigations

Transaction monitoring alerts investigated with the customer's profile and activity gathered and the rationale drafted, and SAR narratives prepared for the cases that escalate.

Sanctions screening review

Name, payment and adverse media screening hits reviewed against the customer's own identifiers, with false positives cleared on a written rationale and possible matches escalated.

Fraud and dispute reviews

Fraud cases, claims and payment disputes reviewed with the evidence gathered and every deadline tracked, and decisions and customer letters drafted for approval.

Third-party risk management

Vendor and fintech partner due diligence read against your standards and the guidance that applies, with gaps, follow-up requests and the assessment drafted for review.

Consumer lending

Loan file and funding review

Loan files indexed, income calculated, conditions and stips checked and dealer funding packages reviewed against the approval, with every finding linked to its page.

Servicing and collections

Borrower calls, texts and emails handled within your procedures and the law, with payments and arrangements set up and everything else handed to your staff.

Borrower disputes and complaints

Credit reporting disputes, mortgage notices of error, payment chargebacks and complaints investigated from the account record and answered before their deadlines.

Total-loss and GAP claims

Claims on financed vehicles worked from the notice of loss to the lien release, with valuations checked, GAP claims filed and add-on refunds collected.

Servicing compliance review

Every servicing and collections call, message and account event checked against federal, state and your own rules, with complaints captured and findings routed for remediation.

Commercial lending

Financial spreading and credit memos

Financial statements, tax returns and property operating statements spread into your template, and the credit memo drafted from the spread, the request and the relationship.

CRE underwriting documents

Rent rolls, operating statements and leases read into your underwriting model, and appraisals and third-party reports checked against your requirements before closing.

Construction draw review

Draw requests checked against the budget, the inspector's report, invoices and lien waivers, with every exception listed before funds are released.

Covenant and portfolio monitoring

Compliance certificates, borrowing bases, borrower financials and property reports read as they arrive, covenants tested and annual reviews drafted for the credit team.

Commercial loan closing

Closing checklists tracked and loan documents compared with the credit approval; title, survey, insurance and entity documents reviewed against your requirements.

Other systems we build

Exam and audit requests

Requests from examiners and auditors answered from your policies, procedures and records, with every response drafted for the responsible manager and linked to its source.

Policy and procedure assistant

An assistant that answers staff questions from the bank's own policies and procedures and shows the source of every answer, so compliance fields fewer repeat questions.

Regulatory change review

New rules and guidance read against your policies and procedures, with each gap listed and a proposed change drafted for your compliance team to decide on.

These are examples of what we can build for banks. Pick one up as-is, have it tailored to you or suggest something entirely new.

Talk to an engineer

Where to start

For most banks we recommend starting with KYC refreshes or alert investigations, because queues are large and past decisions are documented to measure against. Banks with large consumer loan books might start with call compliance review or total-loss recovery instead, and commercial banks with spreading.

The 4-week Opportunity Evaluation tests that choice on your own files before anything is built. It also produces the evaluation set and documentation your validation and compliance teams need, so their review can start before the build.

What the evaluation includes

Questions and answers for banks

How does a language model system get through model risk management?: It is treated like any other model, with a documented design, an evaluation set from your own cases and accuracy monitored before and after launch. We plan for your validation team to review the design before launch rather than after.
Will the system make decisions on customers, alerts or credit?: Decisions on customers, alerts, sanctions hits and credit stay with your analysts, investigators and credit officers. The systems gather, read and draft, and every recommendation carries its evidence.
Do we have to replace our core, case management or loan systems?: The systems work with the core, case management and loan systems you already run, and the evaluation confirms how before anything is built.
How does this fit our third-party risk management?: We go through your vendor due diligence like any other provider. Because the systems run in your environment and you own the code outright, no vendor holds customer data once the work is finished, which usually simplifies the review.

Key terms

A senior team that has done this before

We are a small, entirely senior team with fifty years of combined AI engineering, multiple granted patents and Y Combinator founders, from Yale, MIT, McKinsey, Goldman Sachs, Microsoft and Vista Equity Partners. The people who scope your work are the people who write the code.

Selected work

We built the AI transformation roadmap for a top-50 P&C brokerage, and it is saving the firm millions of dollars in vendor fees it did not need to pay.

Clients include some of the largest banks, RIAs, fintechs and AI labs, as well as a top-3 P&C carrier, a top-10 P&C brokerage, a top-5 trucking brokerage, a leading specialty MGA and a leading trucking MGA.

We run only 3 to 5 engagements at a time, so that every client works directly with the senior team.

About the team

Dozens: of insurance companies, RIAs, banks and fintechs we have worked with
50 yrs: combined AI engineering experience
10+ Patents: granted to members of the team
100%: of the IP we build is assigned to you

Tell us which workflow costs your team the most time.

A 30-minute call with an engineer usually shows whether something is worth building. Within a week you will have a free written point of view on the two or three opportunities we think are real and what each is worth.

Talk to an engineer Other industries we work in