The thing that struck me as I started comparing notes across companies was how identical the stack is. Different teams, different industries, different starting infrastructure — and yet the architecture converges to the same handful of components.
It is always some version of: a coding CLI (Claude Code, Codex, Cursor, or a custom wrapper) running inside sandboxed VMs, hooked into the data warehouse, production logs, and a growing set of internal tools through MCP or homegrown adapters. A Slack interface for kicking off jobs and reviewing output. Some attempt at context engineering — markdown files in a git repo, indexed for retrieval, version-controlled like a codebase. A deployment pipeline. An evaluation harness. A half-built observability layer.
That convergence is the productization tell. When fifty different companies independently build the same thing, the thing wants to be a product. It just has not been one yet.
What the companies building these systems are quietly taking for granted is the decade of platform engineering that made it possible. Shopify did not "stand up an agent platform" — they had well-configured sandboxes, sophisticated context plumbing, internal tool registries, and evaluation frameworks built up over years. The agent layer is the new part. Everything underneath is old infrastructure outsiders do not see.
I've argued elsewhere that the framework cannot be the product — the durable revenue layer is the indispensable infrastructure piece. But the prior question is what the durable architecture even looks like. The stack has settled. The product layer is up for grabs.
Sources
Related Essays
The Atomic Agent Mesh: Architecture, Build-vs-Buy, and the Review Layer
Enterprise AI will not be one mega-agent. It will be a mesh of atomic, auditable units, and the companies that nail review and context will own the next infrastructure layer.
The Agent Infra Maturity Gradient
Mature engineering orgs reuse existing dev infra. Less mature orgs buy off the shelf. Scrappy teams hand-roll everything. The opportunity sits in the gap between them.
The Mesh, Not the Monolith
One mega-agent that handles everything is exhilarating to demo and chaotic in production. Enterprise wants a mesh of specialized agents with human pilots.
Key takeaways
- Internal agent platforms across industries converge on the same architecture — coding CLI, sandboxed VMs, MCP tools, Slack interface, eval harness, observability layer.
- Convergence at this scale is a strong product signal. When fifty teams independently build the same thing, the thing wants to be a product.
- The agent layer rides on a decade of platform engineering most outsiders never see, which is why companies without that base will struggle to replicate it.
FAQ
What does the typical homegrown agent stack look like?
A coding CLI like Claude Code or Codex running inside sandboxed VMs, hooked into the data warehouse and internal tools through MCP, with a Slack interface, a context-engineering layer in git, and a half-built evaluation and observability harness on top.
Why does architectural convergence matter strategically?
Convergence is the productization tell. When companies in completely different industries arrive at identical architectures without coordinating, the design has been validated by the market. The remaining question is who packages it.