Credit proved the model, fraud and compliance followed

Coast started on Oscilar with credit limit decisions and case management, where the value showed up immediately. Manual review time fell from two hours per analyst per day to under 30 minutes, a 75% reduction that returned roughly 750 hours a year to the team at its size.

"We’ve been able to move away from building hundred-line, complex queries, and can now maintain all of our fraud rules within Oscilar."

Delia Nuñez

Director of Risk and Analytics

Coast

Just as important, the risk team could manage more of the logic itself. “We’ve been able to move away from building hundred-line, complex queries, and can now maintain all of our fraud rules within Oscilar,” Nuñez said.

That early experience changed how Nuñez thought about the platform. "I said this very early on when we partnered with Oscilar: I feel like you can use this for everything."

What she saw was a reusable decision layer, not a point solution. Coast could bring in its own data and third-party signals, define new attributes, and apply different logic without standing up another system each time. Credit proved the model first. Fraud followed, then transaction monitoring, sanctions screening and AML, each running on the same customer context.

When fraud changes, Coast can adapt the same day

Three years later, one of the biggest differences is how quickly Coast can respond when a new pattern emerges.

“We've had fraud ring attacks. We've had instances where we checked revenue but then saw credit bust-out behavior,” Nuñez said. “It doesn't take us months, weeks, days. Same day we can go in and tweak.”

Coast's analysts can make those changes themselves. They can test new logic against historical outcomes, see the effect on false positives, inspect the signals behind a spike or miss, and revert a change before it creates problems. AI-assisted workflow building makes it faster to turn an idea into something the team can test.

That gives Coast speed without giving up control. As fraud changes, the team can adjust its decisioning the same day and validate the impact before the new logic goes live.

Scaling the book without scaling the work

As Coast grew, the risk team used automation to keep more of that growth from turning into manual work. “Being able to fine-tune rules and fine-tune where we focus has allowed our team to stay extremely lean,” Nuñez said. “We can very quickly see what's triggering good versus bad cases: where are we wasting our time versus what's effective, how can we automate. We've been able to scale tremendously while still keeping the team lean and functional.”

"We've been able to scale tremendously while still keeping the team lean and functional.”

Delia Nuñez

Director of Risk and Analytics

Coast

Much of that leverage comes from stopping unnecessary work before it reaches an analyst. Cases can be suppressed when they do not warrant review, actioned automatically when the evidence is sufficient, and carried forward with context from previous cases on the same fleet.

Nuñez saw the contrast while interviewing a candidate whose previous job was largely to review bank statements by hand. “I thought, that's insane,” she said. “We rely on a very strong signal without having a single person go in and review bank statements.”

For Coast, that is the value of building automation into the decision layer rather than around it. Case management, suppression and end-to-end workflows have become infrastructure the team can keep extending as the portfolio grows, instead of recreating the same operational work at a larger scale.

Data partners plug seamlessly into the same decisioning workflow

Adding a new data partner does not have to create another system or workflow. Inscribe is the clearest example.

Coast had used Inscribe, an agentic document fraud detection platform, for years to analyze bank statements for signs of tampering and extract credit-relevant attributes, but that step sat outside Oscilar. In 2026, Coast brought it into the existing case flow. Now an analyst can request documents from the case, Oscilar sends the request through Inscribe, and the resulting attributes flow straight back into the decision. Cases that meet Coast’s criteria can close without an analyst reopening them.

For Nuñez, the value is not just the integration itself but how easily it becomes part of the existing decision process. "I don't need an engineer," she said. "If you need anything around revenue verification or document fraud checks, it's just part of one flow within Oscilar."

[A companion post on the Oscilar blog tells the Oscilar and Inscribe story in full.]