The Signal — Coding agents are now the dominant application narrative. Anthropic's singular focus proved the winning bet while OpenAI chased video generation and consumer AI. Enterprise adoption is accelerating hard: Salesforce committing ~$300M annually to Anthropic signals the shift is real and capital-backed.
Consensus: Bullish | Conviction: High
What's Moving
- Anthropic's positioning — Viewed as "rocket ship" by Fortune 500 leadership; enterprise confidence compounding faster than public narrative catches up (via @bindureddy, @allin)
- Multi-model agent swarms — Opus 4.7 + GPT 5.5 Thinking + Gemini 3.2 combined for full-stack software delivery; custom SaaS replacement thesis gaining traction (via @bindureddy)
- Claude Code adoption — Seasoned engineers visibly impressed; agentic coding capability gap now obvious to practitioners who aren't "chronically online" (via @svpino)
- Open-weight competitive threat — MiniMax-M2.7 (230B, open) benchmarking at Opus 4.6/GPT-5.4 level at 5% cost; fast inference (440 tok/s) + cheap unlocks new deployment patterns (via @svpino)
- Developer-centric infrastructure — Slack pivoting hard into AI coding; protocols (MCP, AG-UI) standardizing agentic integration; market rewarding singular focus (via @svpino, @allin)
Blind Spot — LLM limitations on continuous, high-dimensional, noisy data remain unaddressed. YLeCun's signal—that LLMs "totally suck" at real-world sensor/time-series problems—gets buried under agent hype. The coding-agent win doesn't solve robotics, autonomous systems, or real-time control. Cost-per-inference still matters at scale; proprietary model pricing is already unsustainable for commodity tasks. Open-weight models will compress margins faster than most insiders admit.
One Actionable Idea — Track which enterprise SaaS vendors aren't announcing Anthropic/agentic partnerships in next 60 days—they're the ones at existential risk.
Sources: @bindureddy (bullish agent swarms, cautious proprietary costs), @svpino (bullish agentic adoption, tracking open-weight), @ylecun (bearish LLM limits), @sama (bullish scale in emerging markets), @allin (bullish Anthropic strategy win)