How much of the software your company pays for do you actually use? David Brain raises an interesting possibility: if you only need 10% of a SaaS platform, could you vibe code that 10% yourself? Ian Barkin connects the question to a striking development: insurance software company Guidewire acknowledging in its annual filing that customers building their own AI-powered alternatives could affect its business. The discussion turn to this: building the tool might be getting easier, but keeping it running is another question. In this founder discussion, Ian and David unpack HubSpot's explanation for its latest layoffs, what rising AI usage means for the economics of legal tech company Harvey, and why businesses may be better off choosing smaller, cheaper models for routine work. They also explore AI-driven personalized pricing and California's No Robo Bosses Act, including a real client example of an AI-generated employee performance report that appeared convincing until humans checked the details. About the Hosts Ian Barkin and David Brain are co-founders and co-CEOs of magentIQ. Drawing on decades of experience in automation, outsourcing, and operational transformation, they explore where AI creates value, what it actually costs, and why human judgment remains essential. Further reading from magentIQ Explore two of the ideas behind this week's conversation: how to keep human thinking sharp while using AI, and why building your business around a single AI provider carries risks. • What Cognitive Offloading Is, and How to Keep Your Team Sharp While Using AI: https://bemagentiq.com/blog/ • Don't Build Your Business on a Borrowed Brain: https://bemagentiq.com/dont-build-your-business-on-a-borrowed-brain/ Episode chapters and timestamps (2:51) HubSpot cuts nearly 660 jobs and insists AI was not the reason (5:44) From AI washing to operational efficiency, and why the layoffs story is complicated (8:13) Can you vibe code your way out of a Salesforce subscription? (9:02) The 10% SaaS question: how much of your software do you actually use? (11:44) Guidewire warns investors that customers could build their own AI-powered alternatives (13:48) Building your own software is getting easier. What happens when you have to maintain it? (15:42) Harvey's AI economics, and what happens when customers use more tokens than expected (18:13) Using frontier models to get started, then moving the work to smaller, cheaper models (20:45) Why businesses dependent on AI APIs might start switching models more frequently (22:42) From offshoring to model arbitrage: the outsourcing playbook applied to AI (24:35) Why you do not need a Formula One race car for the school run (27:15) RPA 2.0, cheaper decision engines, and choosing the right technology for the task (29:31) Can AI charge you more because it thinks you can afford it? (33:05) Measuring AI utilization, understanding the costs, and getting value for money (36:06) When AI does eight hours of legal work in five minutes, what should it cost? (39:52) California's No Robo Bosses Act, and why humans must oversee employment decisions (40:53) A real client case where an AI-generated performance report reinforced a manager's assumptions (43:27) Why critical thinking matters when AI gives you a convincing answer (47:19) Teaching the next generation when to use AI, and when to think for themselves Listen now, and tell us: if you could build the 10% of your software you actually use, would you? Find us in your favorite streaming service: iTunes: https://podcasts.apple.com/us/podcast/the-magentiq-show/id1896570951 Spotify: https://open.spotify.com/show/033f2oKxnFr5fmcpdSHjSL iHeartRadio: https://iheart.com/podcast/333292440 Got a perspective worth sharing? We are always looking for guests. Reach out at [email protected] (mailto:[email protected]) .
ADVERTISEMENT
