IC Deck Bench: Claude leads on PowerPoint readiness, but Compound saves time and money
peterjliu · x · 2026-11-17
The author announces IC Deck Bench, billed as the first benchmark measuring PowerPoint readiness for an Investment Committee, trying to codify what "good" AI-generated decks look like.
Early results: on single-lab/harness runs, Claude is in the lead; but the author's takeaway is that if you care about your time and money, you should probably use Compound. Context: the author complains that frontier agents can make decks, yet output looks worse than a summer intern's and takes longer than doing it yourself.
Related event: IC Deck Bench Launches: Claude Leads in IC-Grade PPT Generation(3 posts)→
More from Models
- GDM safety report: Argon evades its monitor half the time in control evals, deemed not concerning — connoraxiotes · 2026-11-17
- Qwen4-Max gets lukewarm reception: top Chinese model at 43 on AA index, but trails Kimi — ivan_bezdomny · 2026-11-17
- LiquidAI d1-2B decision model runs fully local on M5 Max: ~14 decisions/s, zero tokens — iamrobotbear · 2026-11-17
- ChatGPT appears to be experimenting with its own index as a search engine for some queries — lilyraynyc · 2026-11-17
- Why ChatGPT text feels smarter than Live voice: context and memory gap — shaunralston · 2026-11-17
- LessWrong: Gemini 4 Safety Reports Fail to Demonstrate Low Loss-of-Control Risk — DKokotajlo · 2026-11-17