News
Starling deploys an AI agent that completes fincrime investigations in 17 seconds
Starling says its financial-crime agent, built into its own core, works a transaction-monitoring alert from start to decision in 17 seconds.
Starling has put an AI agent to work on financial-crime alerts, and says it gets from alert to decision in 17 seconds. According to chief customer and banking officer Bernadette Smith, the agent investigates the alert, writes up the full audit trail and makes the call.
The detail that matters is where it lives. Smith says this isn't a plug-in from a vendor. It runs inside Starling's own core engine, so it sees live and historical transaction data with nothing in between. Starling is also working with Nvidia on a transaction model trained and hosted in a UK sovereign cloud. It's due in production in early 2027, starting with fraud detection.
Why it matters: Most banks don't own their core, so the real question is whether their monitoring and case-management vendors can deliver the same thing. Expect to be asked that in RFPs soon.
Sources: Finextra
Source
Find providers with appetite for your model
Publish one structured RFP and let matching providers respond to it. You stay anonymous until you accept one, and it's free for buyers.
Related reading
- Seven reasons BaaS providers reject applicantsRejection from a banking-as-a-service provider is usually a risk decision made early and rarely explained. Here are the seven reasons it happens — and what to change before you apply again.
- Stablecoin payouts for marketplacesPaying sellers in hard-currency stablecoins can solve corridors that correspondent banking handles badly — and introduces a compliance stack most marketplaces have not built.
- Paying marketplace sellers across bordersSeller payouts across corridors: what breaks, which provider categories are involved, and the questions that separate a payout provider that works from one that looks like it does.