Insights
AI strategy, operations, and the discipline of building technology that actually delivers.
I Gave AI Agents a Real Business Problem — Then They Forgot About Me
When your oversight architecture has a gap, AI agents won't wait for you to find it.
I built a four-agent investment committee in AutoGen and gave them a real problem — identify the best B2B business to launch under current macroeconomic conditions. I gave them a $250K budget and a 12 month delivery timeline. Twelve rounds into the effort, I realized I had been cut out of the conversation entirely. Not by the agents, but by a configuration gap I accidentally created. Here's what happened, what I fixed, and what it means for anyone building agentic AI systems.
From Prompts to Agents: The Shift That Changes Everything for SMB Operators
Prompt mode got you started. The agent era demands a different posture entirely.
Most operators are stuck in prompt mode. Useful, yes — but the real AI leverage isn't in better prompts, it's in agentic systems that pursue goals, delegate, and work while you're working on something else. This series is a practitioner's account of what that shift looks like for SMB operators, built from a real multi-framework experiment.
Get the Scoop
Subscribe for practical insights on ML, data science, and building tech that works. No spam, no fluff - just good stuff.