The 90-minute test that separates real AI products from expensive chatbots
A 90‑minute test helps businesses distinguish truly valuable AI tools from superficial chatbots by focusing on workflow ownership rather than underlying AI.Summary
The article argues that while the underlying AI models (the "building\)) are becoming cheap and commoditized, the real value of an AI product lies in the workflow layer it builds around that model—the shop's ownership of data, integrations, and daily‑use features that make switching costly. It illustrates this with examples such as Cursor, which grew rapidly by owning users' workflow, and contrasts it with logo‑only products that fail because they offer nothing beyond the rented AI. To help businesses avoid buying ineffective AI, the author proposes a 90‑minute test: go off the demo script, trace answer sources, compare output to ChatGPT, test essential features, and check where data and customizations reside after a month. Applying this test to AI tools pitched to Thai SMEs predicts that more than half of those that fail will disappear or be re‑priced within 18 months, emphasizing that the opportunity lies in paying for the workflow, not the underlying AI.
(Source：Nation Thailand)