This is an archive episode, first broadcast in 2024 under the show's previous name. It carries a short synthetic announcer introduction; the conversation underneath it is untouched human audio.
Michael Zipursky is CEO and co-founder of Consulting Success and has advised organisations from startups to billion-dollar corporations across more than seventy-five industries. His observation is that excitement about the technology reliably produces waste, because it pushes people toward building the impressive thing rather than the useful one.
Recorded in November 2024, when the show carried a different name.
What this episode covers
- The big solution trap. Five use cases attempted simultaneously, none of them finished. He describes the shape of it well enough that most people will recognise their own last project.
- Plausible but wrong. Alastair's framing for the genuinely new problem: we have no experience of software that returns confident, well-formed, incorrect answers.
- Build or buy. The two categories - building something custom versus buying a tool that already exists - and how to tell which one you are actually in before you commit budget to either. Most of the waste he describes comes from getting this backwards.
- The AI sales rep pitch. Zipursky was sold a tool for automated outreach and asked the question the vendor had not: what does this do to the brand if it goes wrong at scale? His summary of the offer is blunt.
- AI talking to AI. Automated applications meeting automated screening, and a recruitment loop that no longer serves anyone in it.
- Choosing tools without chasing every release. A practical filter for a market that ships something new every week, and permission to ignore most of it.
- Hallucination, fact-checking and why experts still matter. His argument that subject-matter expertise becomes more valuable rather than less, because someone has to notice when the output is confidently wrong.
- Using AI to find your own use cases. Rather than reading someone else's list and hoping it transfers.
- Agents and an API-first future. Where he thought this was heading, including AI-optimised websites and what that does to how anyone finds you.
- Learning in public. His answer to the unknown unknowns problem, and why he thinks visible experimentation beats private study.
- Two action items. He closes with two things you can do this week, deliberately small.
About the guest
Michael Zipursky is CEO and co-founder of Consulting Success, host of the Consulting Success podcast, and has been featured in Forbes, Entrepreneur and the Financial Post. He has advised organisations ranging from startups to billion-dollar corporations across more than seventy-five industries, which is the vantage point this episode is delivered from: he has watched a great many companies make the same expensive mistake in sequence.
Find him at ConsultingSuccess.com, the podcast wherever you listen, and on LinkedIn - where he asks that you mention you heard this episode.
Resources mentioned
- Consulting Success - his company, podcast and YouTube channel
- Blockbuster, Kodak and BlackBerry - his three examples of organisations that saw the shift and did not move
- Learning in public - his answer to the unknown unknowns problem
Quotes
"Can you help me spam people better?"
Michael Zipursky, on the AI sales tool pitched to him
"We're not used to using software that can give us plausible but wrong answers."
Alastair McDermott
Books - the Human-First AI Series
- Hello! I'm New Here
- An Absolute Beginner's Guide to AI
- Use AI, Stay Human
- The AI Pilot Handbook
- When AI Becomes Your Responsibility
Connect with Alastair on LinkedIn.