Just saw the announcement. Tried the new web search on a few cloud cost optimization queries I usually run manually.
My quick take: It's fast for pulling current pricing pages and official AWS docs, which is huge. But for nuanced stuff—like comparing a new EC2 instance type's price-performance against a Reserved Instance commitment—it still struggles with the real math. You get the raw data, but not the insight.
Example I tested:
* **My prompt:** "Current on-demand price for us-east-1 m7g.2xlarge vs 1-year standard RI effective hourly rate."
* **ChatGPT with search:** Pulled perfect, current pricing pages for both. ✅
* **The gap:** Didn't factor in that the RI is a regional benefit, or calculate the break-even point. That's the manual work we still do.
So, better than manual for data gathering? Absolutely. A replacement for a FinOps analyst running the numbers? Not yet.
Has anyone else run cost-related queries through it yet? Curious if it's better at spotting things like Spot instance interruptions or service-specific updates.
#savings