← back
How to Leverage Domain Expertise — Chris Lovejoy, Notius Labs
Takeaway
Choose how domain experts shape evaluation and improvement, rather than treating their knowledge as occasional advice.
Summary
- Chris Lovejoy argues that vertical AI differentiation depends on operationalizing domain judgment and understanding the last mile of customer workflows.
- Common mistakes include bringing experts in too late, choosing the wrong expertise, or giving experts an ineffective organizational role.
- The oracle model has experts inspect outputs and directly improve prompts, documents, or tools.
- The evaluator model turns expert judgment into metrics and review systems, while the architect model designs feedback systems that improve through usage.
domain-expertisevertical-aiorganization-design
Original description
Granola's first employee was a writer who still reviews meeting note outputs and tweaks prompts directly. Chris Lovejoy says that is not a gap in the org chart. There is no objectively perfect meeting note, so you need someone with taste doing both the assessment and the improvement. He frames this as one of three patterns: the Oracle owns the full loop, the Evaluator defines quality and measures it while engineers improve, the Architect builds systems that improve from usage automatically. Three case studies cover when each is appropriate, what skills to hire for, and what happens when you bring in a domain expert but give them no ownership. Speaker info: https://x.com/ChrisLovejoy_ / dr-christopher-lovejoy https://github.com/chris-lovejoy