The threat nobody is pricing in
In 12 months, a single engineer with the right agent stack will out-iterate most product teams. The teams that survive aren't the biggest — they're the ones whose know-how is documented as the work happens, and whose infrastructure is agent-ready so the next pivot is a prompt instead of a rebuild. I help your platform team build both, in parallel with the shipping you already have to do.
Document the know-how
Decisions, runbooks, and tribal knowledge captured by the same agents that do the work — not a quarterly Confluence audit.
Make infra agent-ready
MCP-native internal tools. Next pivot is a prompt away, not a rebuild.
I already run this stack
Four generic crews end-to-end on hardware I bought myself. I know how it breaks before it breaks for you.
Speed is the moat
Your customers will pick the model that's smartest plus the platform that ships features fastest around it. The first half you already lead. The second half — Claude Code, the agent SDK, MCP server adoption — is platform work that compounds: every well-worn primitive lowers the bar for the next customer who would otherwise build their own loop poorly. The wedge is whoever ships those primitives weeks faster, and the headcount on the other side of the bet doesn't matter if the loop is run by one engineer who already does this end-to-end.
What your competition is doing
Anthropic already invented MCP. The peers below didn't — they're racing to commoditize the platform around models, and each is a useful upper-bound for what 2026 platform work should look like outside your walls.
OpenAI
Shipped Agent SDK + Apps SDK — bet enterprise distribution on platform primitives, not raw model wins.
Direct competitor mirroring Anthropic's playbook one quarter behind. The platform features either ship faster from your side or the default 'agent platform' becomes someone else's stack.
Source: platform.openai.com
Hugging Face
Built the platform layer (Inference Endpoints, Spaces, Hub) into the moat — not the models.
HF's leverage is platform-on-top-of-everyone's-models, including yours. The same shape of work pays back inside Anthropic — every primitive that lowers the bar for an enterprise customer pays compound interest.
Source: huggingface.co/blog
LangChain / LangGraph
Took agent orchestration to enterprise as a platform — agnostic to which model you call underneath.
Useful counter-position: LangGraph customers run on Claude today, but the orchestration layer is what gets sticky. Owning the layer above the SDK (not just the SDK) is the platform-team move worth shipping.
Source: blog.langchain.dev
None of these win on model alone. Same will be true for Anthropic — the platform team is where the next 12 months of customer retention is fought.
Closest match — agent platform
I've shipped the boring infra around interesting agents (loops, retries, evals, observability). That's the work that turns a smart model into a deployable platform — and it's the gap I'd close at Anthropic.
Self-hosted on a 2-node cluster I bought myself: CrewAI crews, Hermes orchestrator, OpenHands autonomous coder, custom MCP servers, Langfuse traces, Postgres + Qdrant, the lot. Four generic crews (architecture, development, marketing, hr) pluggable into different "companies" — the pattern your enterprise customers will need to wire up around Claude Code + the agent SDK.
What I'd do in the first 90 days
First wedge: a per-vertical reference MCP server library — payroll, CRM, ticketing, observability — so the gap between "Claude Code can do anything" and "Claude Code can do MY thing" closes by default for a long-tail enterprise customer. Second wedge: an eval + trace harness that runs on every Claude release so customers can defend "Claude is enough" with cost + quality data, not vibes. Third wedge: pick the highest-friction enterprise onboarding flow and rebuild it as an agent loop that handles 70% end-to-end with a clean human handoff. The pattern, not the feature, is what compounds — and it's portable to every customer after the first one.
If after a short call it isn't the right fit, no pressure. Either way you get the analysis I already wrote about your stack.
Fidel Perez · Agent-platform engineer · 11+ yrs · MCP-native