What I’ve built

These run every day at BrightLeaf Giving and BrightLeaf Digital. None of them came out of a box. I designed them and wired them together.

A shared memory for the company

Rowan is our institutional memory. Every AI tool on the team reads from it and writes to it, so what one conversation learns, the next one knows. It runs on a database with vector search, anyone on the team can ask it about our companies from WhatsApp, and when someone corrects it, the correction sticks.

Voice receptionists

AI agents answer the phone lines for BrightLeaf Giving and BrightLeaf Digital. They answer questions from our documentation and pricing, send real support issues to the right form, take down leads, and book calls on the calendar. We’re also testing outbound calls that get busy people to commit to a demo time.

Automations that do the routine work

A morning digest of what needs my attention. Weekly reviews of our task lists and our search traffic. A queue of social posts that waits for one tap of approval. Most of it runs on n8n and Claude, and all of it reports back to a person before anything important goes out.

What coaching looks like

We start with your actual work, not a tour of tools. Where does your time go, what keeps getting dropped, what would you build if building were cheap? Then we build one real thing together, and you learn the method by using it.

The first thing I teach is LEAD: list, eliminate, automate, delegate. List everything you do. Eliminate what doesn’t need doing. Automate what repeats. Delegate the rest, to people or to agents. Everything else builds on that.

AI is not infallible. I’ve never had a conversation with one where I didn’t have to correct it in some way. A good part of the coaching is learning where to trust it, where to check it, and how to set things up so its mistakes are cheap.