The biggest opportunities can hide inside painful, overlooked work. "That task took me 900 hours and 18 months to settle the affairs of her mom," Mysoor said. That's roughly 12 hours a week for a year and a half, and most families have no idea what’s coming until it happens. 

AI carries the paperwork; people carry the relationship. "We have a human interface… All of that is powered by AI in the back office," Mysoor said. Families work with people while AI absorbs the administrative weight underneath.

Inheritance is a financial access issue. Mysoor framed inheritance as a meaningful source of financial stability and a path to broadening who gets to build generational wealth, well beyond wealthy households.

Estate settlement decides who keeps the assets. Mysoor said 70% to 90% of heirs fire their parents' financial advisor, which makes this the moment institutions either earn the next generation's trust or lose the relationship entirely.

Leaders need to design from a blank page. "I really believe that people need to think bigger," she said, urging the room to imagine what AI makes possible beyond the workflows they already run.

AI is reshaping how legal work gets done. Inanoglu Ozdemir described AI accelerating research, document review, and drafting while improving precision and quality along the way.

Adoption is an organizational project. Product performance matters, and so do committed leaders, practice-area champions, training, and redesigned workflows.

A trusted AI product needs layers around the model. "We think of models as the engine… and Legora as the fancy race car that is built around the engine," Inanoglu Ozdemir said. She described Legora as an agentic operating system with layers for trust, firm-specific context, legal sources, tools, and human-agent collaboration.

Standardize what's common, and let customers shape what's unique. Legora builds shared capabilities into its product while letting legal teams create firm-specific "skills." Inanoglu Ozdemir said customer knowledge stays above the model layer, where clients can inspect and own it.

Pricing should track the work agents do. Legora moved from seat-based to consumption-based pricing to better align price with the work its agents perform.

Leaders are planning on shorter horizons. "The biggest change is I learned to hold plans a bit more loosely," Inanoglu Ozdemir said.

AI raises the bar for expertise. "We all are working harder than we've ever worked before post-AI," Taylor said, pointing to people taking on harder, higher-value work as routine tasks get automated.

Everyone is a builder now. Rho built an internal AI harness that lets nontechnical employees describe what they want in natural language; the tool builds it, and engineering does a lightweight review before it goes live. "You have to have processes, you have to have guardrails, but they have to be super super light," Johnson-Chyzhykov said.

Judgment scales through shared standards. Loza said developing judgment starts with defining what good looks like, which is why Gradient Labs designed its platform so engineers, product managers, and content reviewers can all collaborate on test cases and evals from wherever they already work.

Reliability comes from narrowing the agent's job. For financial workflows, Loza described combining deterministic and non-deterministic systems, breaking agent work into the smallest, most specific tasks possible, and wrapping each in guardrails and evals. Her rule: "let AI do just what it has to do."

Humans stay in the loop. Rho built internal copilots for its operations and customer service teams, plus a Slack lookup tool, with an operations team member giving the AI's answers a light review.

Customers expect AI to be right every time. Johnson-Chyzhykov said consumers' experiences with tools like ChatGPT have raised their expectations for AI inside financial products. "There's zero margin for error," she said.