Site description.
A plain-English guide to AI in marketing for executives and VPs. The five jobs AI does well today, the three traps that catch smart teams, a one-page rule sheet, and a 30-day plan you can start Monday.
AI agents can now do real work on their own, which means they can also make real mistakes on their own. Six simple rules, one story of what good looks like, and a seven-question checklist any executive can bring to Monday's meeting. No tech words required.
A new InstallMap study of 5,630 YC company domains found that a third of startups with an MCP server have no public API at all. What you gain by going MCP-first, what you quietly give up, and how we think about it while building ANCI AI.
AI agents are the new digital insiders: workers that act on their own, talk to each other, and already misbehave at 8 in 10 companies. A plain-language guide for executives on the five new risks and the seven questions that keep them in check.
Building ANCI AI's brain taught us where an LLM belongs in an agent and where it doesn't. Hallucination cascades and runaway token costs pushed us toward a design where the model only extracts intent into a structured Action DSL, and deterministic code does the rest.
Luck looks like chance, but we play a bigger role in it than it seems. A look at blind luck versus controlled luck, personal agency, the 1% rule from Atomic Habits, and Richard Wiseman's newspaper experiment on why feeling lucky helps you notice opportunities.
AI agents are only useful if people can actually interact with them. A look at connecting an AI scheduling assistant to Microsoft Teams - the adapter service that bridges the two systems, progress updates with Adaptive Cards, and keeping conversation state so "tomorrow" means what the user meant.
I had the skills. Unfortunately, I also had to prove I had personality. A short story from 2032: an AI agent gets coached by an AI, interviewed by an AI, and hired by an AI — and discovers that the interviewer might become its boss, its employee, or both.
Vertical AI agents are starting to run the job search on both sides — career agents that find and submit applications, and recruiting agents that screen them. When both the candidate and the company are represented by AI, the resume stops being the main thing, and the real question becomes what these agents are optimizing for.
As AI becomes more capable, having the right context is becoming just as important. Why the next AI race may not be about building smarter models, but about building AI that understands the data, workflows, and rules of the business it works in — and why vertical, specialized agents may win.