Insurance and asset management
Carriers and asset managers have AI pilots in intake, claims and client service that stall in compliance review. We build the gateway, the decision log and the review pack that get them into production, with a person signing what matters.
An insurer does not need AI that decides for it. It needs AI that shows its work, so a person can decide faster.
Where you are
Intake, claims and service teams are trying assistants that read documents and draft the next step. Asset managers want advisors and client service teams to search research, fund documents and client records in one place. The platform is chosen. The compliance review is where the work waits.
State AI rules
A growing number of states have adopted the NAIC model bulletin on insurers' use of AI. It asks for a written program, governance, oversight of third-party models and data, and testing for unfair discrimination. New York has its own guidance on AI and external consumer data in underwriting and pricing.
What waiting costs: an exam question you answer with a pilot nobody documented.
Explainable decisions
When a model shapes a price, a claim or a declined application, someone has to explain the result in words the customer and the regulator understand. A model with no record of its inputs cannot.
What waiting costs: adverse decisions you cannot defend.
Client data
Investment advisers and funds hold client records under the SEC's Regulation S-P, which now requires an incident response program and prompt notice when customer data is exposed. An assistant that ignores entitlements turns one search into a breach.
What waiting costs: a notice letter to clients, caused by a tool built to serve them.
| Market | What supervisors are looking at | What it means for an agent |
|---|---|---|
| United States, insurers | State insurance departments, many acting on the NAIC model bulletin on the use of AI systems by insurers. The New York DFS Insurance Circular Letter No. 7 (2024) on AI and external consumer data in underwriting and pricing. | Every model has an owner, a purpose and a test for unfair discrimination. Third-party models and data are governed like any vendor. Any decision a customer could question can be rebuilt with its inputs. |
| United States, asset managers | The SEC's amended Regulation S-P for investment advisers and funds: an incident response program, oversight of service providers, and notice to customers when their data is exposed. | An assistant searches only what each user is entitled to see. Every query is logged. A vendor incident reaches you in time to tell clients. |
| Other markets | Each market has its own insurance and securities supervisors. | The questions are the same and the filings differ. We confirm the local rulebook at the start of every engagement. |
What is in the way
Insurance runs on documents, and the documents sit in claims, policy and imaging systems that were never built to be read by a model. Add a third-party model and a state that wants it tested, and the pilot needs a control layer it was never given.
If an examiner picked one claim your AI touched last month, could you show what it read, what it suggested and who decided?
One claim, two files
Without
The decision is there. The reason is not.
With
One record. The reason is written down.
How we clear it
We put the documents, the controls and the log in place first, then the agent. A person signs anything that changes a price, a claim or a client's account.
Agents on Gemini Enterprise, with Document AI reading submissions, claims and medical records, checking completeness and routing the case. An adjuster or underwriter signs.
You hold one agent in production in first notice of loss or new-business intake, with a decision log.
Gemini Enterprise and agentsGemini Enterprise grounded on research, fund documents and client records, with each person's existing access enforced on every search.
You hold an assistant advisors use that shows each person only what they may see.
Gemini Enterprise and agentsApigee in front of policy administration and third-party administrator systems, with policy, client and fund data governed on BigQuery.
You hold one API layer and one governed data set your compliance team can map.
Apigee and APIsA model inventory, third-party oversight records, test results your actuaries can review, and a sample decision log.
You hold the written AI program file your regulator expects, ready before the exam.
Security and responsible AIEasy to start
Choose one workflow, name one owner, and give us time with the people who sign it off.
We bring engineers and the gateway design, we work inside your environment under your security rules, and we write the review pack alongside the build.
A good first workflow for a carrier is first notice of loss or new-business intake. For an asset manager it is an advisor assistant on approved research. See the fixed-price ways to start.
What the work feels like
One workflow is chosen and one person owns it. The questions your reviewers will ask are written down before any build starts. Your second line sees the controls in the first week, not at the end.
Governance goes in before the first agent does: the gateway, the audit trail and the access rules. Your compliance team sees the decision log working in the first sprint demo.
A working demo you can show your own review board, not a status slide.
A decision log that can rebuild any action, and a go-live date that was agreed at the start.
Seen it before
Production readiness check
A short readiness check on a live or planned agent workflow. You get a score and the principles worth another look.
Take the readiness checkThat depends on your states. Many have adopted it and others act on their own rules. We build the inventory, governance records and test evidence the bulletin asks for, and your compliance team maps them to your states.
No. It reads, checks and drafts. An adjuster or underwriter decides, and the log records both.
The assistant checks each person's existing entitlements before it searches, and logs every query. It does not keep its own copy of client records.
Yes. The controls are the same. The documents differ, from claim photos to medical records and beneficiary forms.
In your environment, under your keys and perimeter. Agents read governed data in place instead of copies of it.
Tell us what you are working on. Our Customer Partnership Team will reply to arrange a conversation.
Get answers to your questions
AI assistant. It can make mistakes, so check anything important with our team.
