The economic world model your product or agent builds on.

Turn the forces that move markets — energy, freight, feedstock, capacity, FX, macro — into confident, defensible decisions in real time. Call it over API or MCP, and ship decision-grade reasoning into your product or agent without assembling and maintaining your own.

mcp.sybilion.dev/mcp API-first Real-time inference Explainable by default

Two ways to build on it.

Build on it · API

For deterministic systems you operate.

Quant and trading desks, and product, data and planning teams OEM-ing external-market intelligence — decision-grade signal as better inputs, with no workflow change.

Build on it · MCP

For runtime agents that call it as a tool.

AI-native teams building vertical procurement, trading and supply agents — the external-world grounding a foundation model can't provide, called the moment a decision is made.

Build on it — don’t rebuild it.

Today every product and agent assembles its own partial, lagging picture of the market — stitched from feeds, models and reports, rebuilt by every team, and rarely ready the instant a decision has to be made. Build on one shared model instead.

“We were rebuilding the same market logic in three tools. Now they all decide from one model.”

Call it like a tool. Get a decision, not a data dump.

One request returns a decision-grade answer: the verdict, the modelled impact, the evidence and the forces behind it — ready for a system to act on or an agent to reason from.

// MCP — your agent calls Sybilion as a tool
sybilion.decide({
  question: "Supplier wants +8% on propylene,
             USA. Is it justified?",
})
 
→ {
    verdict: "renegotiate",
    modelled_impact: "+2.1%",
    target: "€91–96/t",
    evidence_strength: 0.92,
    drivers: ["energy", "freight", "feedstock"],
  }