Teams that are technically proficient are already building their own productivity harnesses. They use Claude Code to spin up a standup skill, wire it to five MCP servers, and generate a daily summary. It works — until it does not.
The failure mode is predictable. Nothing is cached, so every run makes massive token calls. It is slow. Accuracy degrades when one data source times out. The agent never references the last time it ran the skill — it starts from scratch every execution. There is no learning, no feedback incorporation, no memory.
This is the infrastructure gap that actually matters. Not another wrapper on top of an LLM. Not another Slack bot. The gap is in the data layer — the context and memory infrastructure that agents need to actually improve over time. It is the kind of annoying, non-mission-aligned plumbing that enterprises will pay someone else to handle rather than build themselves.
No one has won this space yet. There is no dominant player in agent memory and context. The frontier labs have not productized anything meaningful here. And the enterprise need for a managed, secure, cost-effective context layer is not going away. The orchestration framework on top will commoditize. The data layer underneath will not.
I've argued that every obvious agent idea gets cloned in weeks, and the corollary is that the durable moat is not orchestration cleverness. The moat is the boring plumbing that lets agents remember, reuse, and learn. Build there.
Sources
Related Essays
Agent Memory Is Unsolved. Workflow-Scoped Learning Is Not
General-purpose agent memory is still a research problem. The opportunity is workflow-scoped learning that compounds — pick the constraint, and the memory problem stops being intractable.
Every Obvious AI Idea Gets Commoditized in Weeks
The window between novel agent product and commoditized feature has compressed from years to weeks. Founders who pivot fast lose. Founders who stay in a hard problem compound.
Tech Context Is Tractable. Org Context Is Not
The hardest unsolved problem in agent infrastructure is not compute or sandboxing. It is context — and most of that context lives in people, not repos.
Key takeaways
- Homegrown productivity harnesses fail because nothing is cached, accuracy degrades on timeouts, and agents start from scratch every run.
- The infrastructure gap that matters is in the data layer — managed, secure, cost-effective context and memory for agents.
- The orchestration framework on top will commoditize. The data layer underneath will not.
FAQ
Why do homegrown agent harnesses break in production?
Nothing is cached, so every run makes massive token calls. It is slow. Accuracy degrades when one data source times out. The agent never references the last time it ran the skill — it starts from scratch every execution. There is no learning, no memory, no improvement over time.
What is the defensible infrastructure layer for agents?
The context and memory layer underneath agents — the managed, secure, cost-effective plumbing that lets agents improve over time. Enterprises will pay someone else to handle this rather than build it themselves. It is the kind of annoying, non-mission-aligned work nobody wants to own.