Still chasing 10,000 real connections on ๐.
Who I want to meet this week:
1. Founders running lean with AI agents
2. Recruiters testing AI in hiring
3. Side-project builders
4. Ops people who hate busywork
What are you building? Drop it below ๐
Flying to Pune this morning.
Before I left home, an AI agent checked three of us in, emailed the boarding passes, and started watching the gate. It pinged me once, when the gate changed.
That's the bar for agents: do the boring work, and only speak up when something changes.
@Saurabhksinhaa We caught the risk while reviewing a batch before it went out, so we held it and shipped the guard first. Good to see Employ Labs sweating the same detail.
Shipped a small fix in our recruiting AI this week: a campaign can no longer text the same candidate twice.
Sounds minor. It isn't. One repeat text and a candidate stops trusting the recruiter behind it.
AI in hiring should make you more careful with people, not louder.
Running a team of AI agents taught me the same lesson managing people did.
When something goes wrong, don't just say sorry and move on. Fix it where it can't happen again: a check, a rule, a better tool.
An apology fixes today. A guardrail fixes every day after.
@manuseen007 Same here. Every agent has one job, a clear list of what it can touch, and someone it reports to. Without that, you don't have a team, you have a pile of scripts.
My rule for AI agents in ops:
They can read anything, draft anything, and fix anything that's reversible.
Anything that leaves the building, an email, a post, a payment, waits for a human yes.
Fast inside. A gate at the door.
My goal on ๐ is 10,000 real connections.
Looking to connect with:
1. Founders and solo builders
2. Dads building on the side
3. AI-curious people
4. AI agent builders
5. Sales and marketing folks
6. High-agency people
If this sounds like you, say hi below ๐
@mathiasvrb Both. Claude Code on Fable 5.2 writes the code, and my own agents sit on top running ops, reviews and deploys. Code is the easy part now. Orchestration is where it gets interesting.
Made a call today: every line of code across my companies is now written by Claude Fable 5.2. One model, no switching. I'll share what breaks and what holds up.