Build on the strategy substrate.
For engineers, technical leads, and AI-platform teams building agents that need to reason against strategic context.
MCP-native by design. Live in the Anthropic Cowork marketplace. Composable with Anthropic Memory. Built so the strategic ontology is just another query away.
Every agent has the same context problem.
The answer to that question is the strategy. It lives in slide decks, board minutes, off-site notes, and the heads of senior leaders. It has never been structured. It has never been queryable. And so every agent every team builds has to reconstruct it on the fly, every time, from scratch — or simply do without.
Stratafy makes that answer addressable. The substrate persists mission, vision, principles, strategies, initiatives, objectives, metrics, decisions, assumptions, risks, and signals as machine-readable infrastructure. Your agents stop reconstructing context and start reasoning against it.
How Stratafy composes.
Stratafy is MCP-native from launch. Any MCP-aware agent — Claude, ChatGPT, Gemini, Microsoft Copilot, or your own — can query and update the substrate using the standard protocol. No bespoke integration. No SDK lock-in.
Stratafy plugins (genesis, guardian, team, expert-cto, expert-chro) are live in the Anthropic Cowork marketplace today. Install in any Cowork workspace and Claude agents immediately reason against the substrate.
Anthropic's Memory tool stores agent state. Stratafy stores strategic state. The two compose: Memory remembers what happened in a conversation; Stratafy remembers what the organisation is trying to do. Different layers, complementary roles.
Mission, vision, beliefs, principles, strategies, initiatives, objectives, metrics, decisions, assumptions, risks, signals — each a first-class object with versioning, addressability, and structured relationships.
Four ways to build with Stratafy.
Your enterprise agents don't have to reconstruct strategic context every session. An MCP query against a Stratafy workspace returns the active strategies, the objectives, the assumptions each rests on, the metrics, and the signals currently affecting them — all structured, all queryable, all consistent across sessions.
If you're working inside an advisory firm or consulting org, your firm's methodology doesn't have to live as a deck. Build it as a Stratafy Expert plugin — a runtime that lives alongside the client's strategic substrate and contributes the firm's framework as code, not artefacts.
Any agent stack — CrewAI, LangGraph, AutoGen, Copilot Studio, or your own — can reason against Stratafy substrate via MCP. The substrate becomes the strategic-context layer your existing agent infrastructure queries when it needs to know not just how, but why.
The Stratafy plugin ecosystem is open. Methodology plugins, role-specific experts, integration plugins for vertical use cases — the marketplace surface in Cowork is where they ship and where customers find them.
Want to go deeper on the schema?
The full architecture — twelve entity types, the four-layer stack, the composition model with MCP and Memory — lives on the Substrate page. That's where to go if you want the architectural deep-dive before writing any code.
Read the substrate architectureWhat's live now.
stratafy-genesis, stratafy-guardian, stratafy-team, stratafy-expert-cto, stratafy-expert-chro — available through early access today.
Read and write strategic state via standard MCP protocol. No SDK required.
Coming soon. For now, the existing plugins serve as reference implementations. Reach out for early access to the developer guide.
On the roadmap. If you need to integrate Stratafy with a custom agent stack outside Cowork, get in touch — we're prioritising based on real demand.