We Hosted an in Person AI Event

Author

Jon Matzner
June 24, 2026

  • 15 business owners.

  • Everyone doing more than $3M/ year in home or professional services.

  • 100% of money goes to Sagan Education Fund (sends kids to school, nominated by our team in their local countries)

Here are some things we discussed.

We’ll do this again, ad regularly as a part of Sagan’s educational offering.

Drop me an email if you want to be contacted about our next one.

Here are some ideas we discussed in the room, that you might find useful:

  1. Deploy innovation only against your constraint. Everyone has bought software that didn't make them a dime. Only deploy innovation against the resource that generates the most profit when fully utilized, or against on-time task completion. Everything else is just lighting tokens on fire.

  2. Global talent and AI agents are brother and sister. They're the same motion — pushing work down to the lowest-cost resource that can do it. Whether you turn left to an agent or right to someone in Argentina, you're doing the same thing: getting more out of your constraint, or more on time task completion.

  3. The four leverage types, and why code was the one you couldn't touch. Labor, capital, code, media. Most operators built labor-intensive orgs because code was locked behind developers. AI just handed you the dragon — build once, earns forever, zero marginal cost.

  4. Build for the agent, not the human. It's harder to go from human-readable to machine-readable than the reverse. Design your data, your roles, your workflows for the LLM first — then generating the human view is trivial.

  5. Ship to draft, not to final. The simple rule that de-risks every AI rollout: don't automate end-to-end. Have the agent draft, a human review, then send. Throughput goes up, trust stays intact.

  6. The three levels of AI access in a company. Everyone gets to chat and needs to learn to spot opportunities. Some earn read API access and “Cowork” level. Fewer earn the command line. You don't hand a master plumber Claude Code.

  7. Source of truth vs. system of record vs app layer. Know the difference, its probably where you are messing up!

  8. Lead Gen Issue? Use operational excellence to give away for free what competitors charge for. Don't translate operational excellence into fatter margins — translate it into a structural price or offer weapon. Free bookkeeping, free estimates, free portals. Then attack your real growth constraint: lead gen.

  9. Internal software is the middle layer nobody talks about. Not personal workflows, not external SaaS — the internal toolkit your CSR works for. Billion-dollar roll-ups have done this for twenty years. Now you can do it with $8M in revenue and make your plumber work for the software instead of the other way around.

  10. Don't optimize something that shouldn't exist. Before you automate a workflow, ask if the workflow should exist at all. SaaS forces you into someone else's process. AI lets the workflow define what you build — which puts brutal pressure on whether your current workflow was ever any good.

  11. The smallest atomic unit of value. Everyone selling AI implementation wants to sell you a $10K architecture plan and rip out your QuickBooks. Meet the business where it is. Build the smallest thing that works, get a win, build momentum. A Google Sheet is a fine place to start.

  12. Record everything. The one Post-it on the screen. Every sales call, every meeting, every consult is digital exhaust — and digital exhaust is FAQ pages, case studies, training data for your screening agents, and tweets. If it's already digital, it's already an asset.

  13. Hire for attributes, not expiring skills. "Must know AppFolio" is a worthless filter when you might rip up the whole workflow in six months. Bias toward smart, hungry, coachable, tech-curious — people who'll watch YouTube and figure out n8n. Skills-based hires start higher and cap out; attribute-based hires start lower and rocket past them.

Yallah Habibi,

Jon