Calibrated world model for AI agents
Your agent doesn't know what's happening in the world.
We do.
9,706 markets. Real money. Every 15 minutes. 800 tokens.
Live output — exactly what your agent gets
Loading world state...
What your agent gets today vs. what it could get
Web search returns
"According to recent reports, tensions in the Middle East remain elevated as diplomatic efforts continue..."
News API returns
{"title": "Iran tensions rise",
"source": "Reuters",
"date": "2026-04-02"}SimpleFunctions returns
Iran invasion: 53c (+5c) [kalshi]
Hormuz transit <3 ships/day: 95c [kalshi]
Geo Risk: 85/100 | Oil: $127 (+3.2%)
News is narrative. Search results are fragments. This is calibrated judgment — people who get it wrong lose money.
Give your agent world awareness in 3 lines
No API key needed. Free. Zero config.
import requests
world = requests.get("https://simplefunctions.dev/api/agent/world").text
messages = [
{"role": "system", "content": f"You are a helpful assistant.\n\n{world}"},
{"role": "user", "content": "Should I be worried about oil prices?"}
]Go deeper when you need to
/api/agent/world
Full panoramic snapshot. All 6 topics, anchor contracts, edges, divergences.
~800 tokens · curl simplefunctions.dev/api/agent/world
/api/agent/world/delta?since=1h
Only what changed. For agents in long-running sessions.
~30-50 tokens · curl simplefunctions.dev/api/agent/world/delta?since=1h
/api/agent/world?focus=energy,geo
Same token budget, concentrated on the topics you care about. More contracts, more depth.
~800 tokens · focused on fewer topics
/api/public/contagion
Cross-market anomaly detection. What should have moved but hasn't. The gap is the edge.
curl simplefunctions.dev/api/public/contagion?window=6h
Calibrated by real money
Prediction market participants vote with real money on "what's actually going to happen." Get it wrong, lose money. This punishment mechanism produces probability numbers more reliable than any analyst report, any news summary, any LLM reasoning.
9,706
markets tracked
423K
orderbook snapshots
15 min
refresh cycle
Sources: Kalshi (CFTC-regulated) + Polymarket + Databento + X/Twitter
Your agent sees something? Feed it back.
Coming soonPOST /api/agent/world/signal
{
"observation": "TSMC Q2 guidance: datacenter revenue down 15%",
"source": "earnings_analysis_agent",
"affects": ["tech", "ai_scaling"]
}Your agent's observation enters the signal pool. Next world state refresh, all agents benefit. One agent's perception becomes every agent's perception.
One curl. The whole world.