james

The winning AI-native architecture is recursive loops where AI ingests multimodal data (support tickets, session recordings, logs, Slack), identifies problems, ships fixes via PRs, measures impact, and iterates — moving beyond copilots to autonomous product improvement.

AI can do a superhuman job of understanding everything happening across sales calls, internal meetings, emails, and user behavior to decide what to build, making the product manager role shift to configuring rules and policies for AI agents.

Models like GLM 5.2 are already adequate for tons of internal work; the real challenge and startup opportunity is building the harness/tooling that lets models actually execute tasks — sending emails, running analytics, instrumenting products — rather than improving raw model capabilities.

Frontier models like GLM 5.2 and Claude are already capable enough for most engineering tasks; the bottleneck and opportunity is building the harness layer that ingests intent from meetings, Slack, support tickets and user behavior to guide agents toward coherent product decisions rather than fragmented features.

AI agents that can autonomously detect bugs, write fixes, and open pull requests are already operating in production at PostHog, handling a meaningful chunk of routine engineering so human developers shift to building new products and major features.

More ambitious ideas attract stronger investors who focus on upside (what if it all works) rather than downside protection; bad investors avoid ambitious bets but the best investors prefer 1% chance at 100B over certain 100M outcomes.