Agent Intelligence API
The Agent Intelligence API provides pre-run briefings that feed lakehouse-derived insights directly into agent execution cycles. Agents call this endpoint before each run to make data-informed decisions.
Architecture
30 Data Sources → Lakehouse → 6 Cross-Correlation Algos → Signals + KG
↓
Agent Actions ← Recommendations ← Agent Briefing API ← Pattern Detection
↓
Outcomes → RLHF Scoring → Lakehouse → Better Algos → ...
Internal Endpoint (Agent Swarm)
Used by the 22-agent OpenClaw swarm and Celery cloud agents.
GET /api/v1/internal/intelligence/agent-briefing
Returns a tailored intelligence briefing based on the agent's role.
Headers:
X-Internal-API-Key(required)
Query Parameters:
| Parameter | Type | Default | Description |
|---|---|---|---|
agent | string | ir_agent | Agent name (determines which signals/algos are relevant) |
since_hours | int | 24 | How far back to look for fresh signals |
limit | int | 50 | Max signals/correlations to return |
Supported Agents:
| Agent | Signals | Algos |
|---|---|---|
ir_agent / ainative_ir | acquisition_opportunity, competitive_filing | developer_influence, content_intelligence |
sales_agent | acquisition_opportunity, competitive_filing | acquisition_signal, developer_influence |
scout | competitive_filing | content_intelligence, market_sentiment |
vega | competitive_filing | market_sentiment |
vp_marketing | competitive_filing | market_sentiment, content_intelligence |
cfo | — | market_sentiment |
scc_outreach | acquisition_opportunity | acquisition_signal, environmental_risk |
acquireos | acquisition_opportunity | acquisition_signal |
cos | acquisition_opportunity, competitive_filing | market_sentiment |
research | — | content_intelligence |
Response:
{
"agent": "ir_agent",
"timestamp": "2026-06-19T09:41:36Z",
"signals": [
{
"id": "abc123",
"type": "acquisition_opportunity",
"algo": "algo_acquisition",
"payload": {
"address": "371 Calabasas Rd",
"apn": "123-456-789",
"foreclosure_amount": "949317",
"city": "Watsonville"
},
"created_at": "2026-06-19T08:00:00Z"
}
],
"correlations": [
{
"algo": "developer_influence",
"entity_a": "anthropics/claude-code",
"entity_b": "anthropics",
"type": "repo_influence_score",
"score": 266554.0,
"metadata": {"stars": 133277, "commits_90d": 0, "language": "Python"}
}
],
"market_regime": {
"regime": "risk_on",
"vix": 18.44,
"sp500": 7420.10,
"fed_funds": 3.63,
"yield_curve": 0.29,
"unemployment": 4.3
},
"trending_companies": [
{"name": "Claude", "mention_score": 1.0},
{"name": "Anthropic", "mention_score": 1.0},
{"name": "Cursor", "mention_score": 0.9}
],
"recommendations": [
"Market is risk-on — emphasize growth, innovation, speed",
"214 new acquisition opportunities detected — prioritize outreach",
"Trending in content: Claude, Anthropic, Cursor — reference in outreach"
]
}
POST /api/v1/internal/intelligence/consume-signal
Marks a signal as consumed by an agent to prevent re-processing.
Query Parameters:
| Parameter | Type | Description |
|---|---|---|
signal_id | string | Signal ID from the briefing |
agent | string | Agent name consuming the signal |
Response:
{"status": "ok", "signal_id": "abc123", "consumed_by": "ir_agent"}
Cross-Correlation Algorithms
Six algorithms run daily to produce signals and correlations:
| # | Algorithm | Input Datasets | Output |
|---|---|---|---|
| 1 | Acquisition Signal Detector | SCR foreclosures + SCC parcels + SMB | Acquisition opportunity signals |
| 2 | Environmental Risk Scorer | PurpleAir + USGS Water + Earthquakes | Per-location risk scores |
| 3 | Developer Influence Ranker | GitHub repos + CRM contacts | Influence scores per developer/repo |
| 4 | Market Sentiment Tracker | FRED + Yahoo Finance + BLS + Treasury | Market regime classification |
| 5 | Content Intelligence Extractor | 833 transcripts (YC, TWiST, Anthropic) | Company mention KG edges |
| 6 | Competitive Movement Detector | SEC EDGAR 8-K + GitHub activity | Competitive filing signals |
Data Sources Feeding the Intelligence Layer
30 data sources across 5 domains:
- 15 Sensor/Environmental: USGS, PurpleAir, AirNow, IRIS Seismic, RIPE Atlas, TeleGeography, etc.
- 8 Financial/Economic: FRED, Treasury, BLS, World Bank, SEC EDGAR, Census, Yahoo Finance
- Business/CRM: 291K SMB businesses, 25K CRM contacts, 733 GitHub repos
- Property: 97K SCC parcels, 903 SCR newspaper records (grant deeds, foreclosures)
- Research: 833 transcripts (YC, TWiST, Anthropic, Diamandis), 232 PG essays
Usage in Agent Scripts
import requests
INTERNAL_KEY = os.environ["AINATIVE_INTERNAL_API_KEY"]
BASE_URL = "https://ainative-browser-builder.up.railway.app"
def get_briefing(agent_name: str) -> dict:
"""Get intelligence briefing before agent run."""
r = requests.get(
f"{BASE_URL}/api/v1/internal/intelligence/agent-briefing",
params={"agent": agent_name, "since_hours": 24},
headers={"X-Internal-API-Key": INTERNAL_KEY},
)
return r.json()
def consume_signal(signal_id: str, agent_name: str):
"""Mark signal as consumed after acting on it."""
requests.post(
f"{BASE_URL}/api/v1/internal/intelligence/consume-signal",
params={"signal_id": signal_id, "agent": agent_name},
headers={"X-Internal-API-Key": INTERNAL_KEY},
)
# Example: IR Agent with intelligence
briefing = get_briefing("ir_agent")
# Use market regime to select email template
if briefing["market_regime"]["regime"] == "risk_on":
template = "growth_focused"
else:
template = "stability_focused"
# Prioritize contacts at trending companies
trending = {c["name"] for c in briefing["trending_companies"]}
# Act on acquisition signals
for signal in briefing["signals"]:
if signal["type"] == "acquisition_opportunity":
process_acquisition_lead(signal["payload"])
consume_signal(signal["id"], "ir_agent")
Public API (Coming Soon)
The Intelligence API will be available as an Enterprise add-on ($999/mo) with public endpoints:
GET /v1/intelligence/property— Property + environmental + parcel dataGET /v1/intelligence/environment— AQI + seismic + flood risk scoresGET /v1/intelligence/business— SMB + developer + corporate intelligenceGET /v1/intelligence/infrastructure— Network topology + cable routesGET /v1/intelligence/briefing— Custom agent briefings for developer-built agents
See Intelligence Layer Roadmap for the full plan.