Claude vs Grok : Head to Head!
Because I'm not busy enough. Anyway, I've got a key domain design choice coming up. Its really hard, and the temptation is just to build a database with really about 1,000,000 rows in it to solve the problem. I worked with Grok and with Claude to discuss the problem and different approaches.
Grok went first, and kept avoid making decisions and was referring them back to me ("what's your gut feeling?"). Eventually it settled on a dynamic service that will match together reference data on-demand from an AWS Lambda rather than denormalise into a massive table.
Agent-Sam (Claude) had a different plan, and went all-in on Postgres. He felt that the massive data scale meant that a database was just the right tool. He planned a db structure and did his estimates. I got them both to write up their conclusions, and then I swapped them over.
They both preferred the other's solution. No fighting, no arguments, but not conclusions either :-s
