A One-Paragraph Prompt Won't Build Your Business
Episode eleven: Jason Patel burns a day of tokens on his build system while the product waits. Matt and Jason compare Opus 5 and Fable 5.2, sort useful edge cases from analysis paralysis, and talk AI meeting notes nobody reads. From there: why a company laying off engineers for AI may just be out of ideas, why faster code doesn't close the loop on a business, when replacing a SaaS subscription makes sense, and what maintenance does to the savings. Plus claymation clips, private equity rollups, Matt's search for a business to buy, and whether small software companies need venture funding.
Episode eleven. Jason Patel burns a day of tokens on his build system while the product waits, and he and Matt compare Opus 5 and Fable 5.2.
What we talked about:
- 0:00 Cold open: Then you didn’t have enough ideas
- 0:25 Fable burned a whole day on edge cases
- 2:47 Did Opus 5 get worse this week?
- 5:49 What Fable is actually good for
- 8:57 Is Claude’s temperature turned down?
- 10:12 AI video: stay out of the uncanny valley
- 12:10 What worked for our Shorts
- 13:11 Was the AI revolution oversold?
- 14:04 AI notes nobody reads
- 17:07 Software factories and one-paragraph prompts
- 18:40 If AI is that good, why lay people off?
- 20:18 Replacing your SaaS: Jason does the math
- 23:43 We tried this with outsourcing
- 25:39 When building it yourself is worth it
- 27:14 Is software becoming a commodity?
- 28:16 The MBA-ification of tech
- 29:46 Where’s the Linear for CRMs?
- 32:41 Will only big companies innovate?
- 34:00 Vertical SaaS with two engineers
- 35:45 VC money or a PE roll-up?
- 38:33 Roll-up savings aren’t in the cloud bill
- 42:13 Where’s the AI productivity boom?
- 43:06 Buying a business with an SBA loan
- 44:32 Can you still build an employee-owned company?
- 46:40 TinySeed and MicroConf
Mentioned in this episode: