Building AI trust in regulated markets

​Product orgs in regulated industries have always had to trust something to review their work, whether it's a team of human analysts or a rules engine that blocks any content containing the word "guarantee." But how do you trust a review system powered by AI when it works, as all AI does, probabilistically?

Facilitated by Jay Nathan, Co-founder and CEO at Balboa, and featuring Kirk Sadler, VP of Product at Casepoint, this session looks at how to overcome AI risk one step at a time.

​Kirk has run compliance-critical products, including tools for SEC-compliant advertising review and tools for legal billing review. In his line of work, a false positive is cheap but a missed violation is not, which makes changes in process or technology hard. So most enterprises land in the trap of deploying AI but then just having employees re-check everything it does, which is conservative but also eliminates the efficiencies. But Kirk has helped teams learn when, where, and how it's appropriate to trust the work done by agents, and seen the huge efficiency gains it unlocks.

Together, we'll explore:

• ​How product teams earn customer trust in AI features while mitigating risk for everyone

• ​Breaking a workflow into pieces to find where AI has earned autonomy and where humans should keep reviewing

• ​Knowing when a workflow is ready to climb from full human review to hybrid sampling to full autonomy

• ​Using drift detection to shrink sampling on mature segments and escalate back when accuracy drops

• ​How forward deployed engineers (FDEs) run the experiments, refine prompts, and decide where humans stay in the loop

​This session is made for product leaders, ELM and GRC leaders, and anyone accountable for AI output in a regulated industry who wants to go from "10x costs" to "10x capability" with a data-backed path to autonomy.