Internal AI PlaybookGet early chapters

About / Internal AI Playbook

Field notes from building internal AI at work.

I build internal AI systems at an NYSE-listed company, from product knowledge and AI SDR workflows to customer success, expansion, and revenue automation. I’m writing the playbook I wish I had when I started.

AI advice is easy to publish. Operating knowledge is harder to earn.

The same problems keep showing up: product knowledge has no owner, customer context is fragmented, an agent has too much permission, or nobody can tell whether a workflow changed an outcome. Those are the problems I write about.

I start with the job as people do it today.

  1. Map the work

    Find who does it, what slows them down, and which evidence they trust.

  2. Define the system

    Name the decision, owner, inputs, permissions, and expected result.

  3. Test real cases

    Run missing data, conflicting records, exceptions, and human overrides.

  4. Measure what changed

    Check adoption, decision quality, downstream work, and the business result.

Start with the problem you need to solve.