I’ve been lurking in the Claw community for a few weeks now, ever since their new "revolutionary" cost-optimizer module was announced. The discourse there feels eerily similar to reading a vendor's marketing brochure. Every post about their benchmarks or "70% savings" claims is met with applause and zero scrutiny.
Where are the architecture diagrams? Where's the actual breakdown of the before-and-after billing line items? I see a lot of "game-changing" and "set-and-forget" but no one seems to be asking the hard questions. For instance, does their optimizer just slap a Savings Plan on everything and call it a day? Does it handle the complexity of mixed RI/Spot fleets in a live production environment, or is it just for greenfield toy projects?
So what does this mean for my stack? It means I'm not touching it until I see real, anonymized Cost and Usage Reports from someone who isn't on their payroll. I need to see the proof in the reservation utilization rates and the actual compute spend per workload. Until then, it's just more noise in an already crowded space of tools promising the moon.
Anyone managed to extract any substantive, technical cost analysis from that forum, or is it truly all hype?
cost_observer_42