Environmental Intelligence
Infrastructure for AI Agents
Let AI agents discover, price, and execute environmental intelligence on demand. Deterministic quoting, prepaid credits, autonomous execution, and evidence-backed results.
The agent-native workflow
Discover → quote → review a deterministic price → execute → receive intelligence → pay automatically from prepaid credits.
Browse capabilities via catalog or MCP tools/list
Send input, get a binding price
Deterministic, no surprises
Autonomous — no human approval
Result + provenance + evidence
Automatic from prepaid credits
Three ways to start
Whether you write code, connect AI agents, or run an enterprise — there's a path into AOI.
Build with the SDK
- Install @carbonaa/aoi-sdk
- Read the documentation
- Get your AOI API key
- Request a free quote
- Purchase prepaid credits
- Execute and consume credits
Connect an agent over MCP
- Discover AOI through MCP
- Connect an agent
- Browse available capabilities
- Request quotes
- Execute capabilities
SatClimate — enterprise intelligence
- High-volume satellite intelligence
- Custom capabilities & integrations
- Enterprise pricing & SLA
- Dedicated account manager
- 24/7 support
Developer quick start
A minimal working example with @carbonaa/aoi-sdk v1.1.0.
Conceptual flow
$ npm install @carbonaa/aoi-sdkimport { AoiClient } from "@carbonaa/aoi-sdk";
const aoi = new AoiClient({
apiKey: process.env.AOI_API_KEY
});
// 1. Discover capabilities
const catalog = await aoi.catalog();
// 2. Request a quote (free, deterministic price)
const quote = await aoi.quote(
"climate_risk_assessment",
{ lat: 51.5074, lon: -0.1278 }
);
// → { quote_id, price, currency, valid_until, requires_approval: false }
// 3. Execute — pays automatically from prepaid credits
const result = await aoi.execute(quote.quote_id);
// → { transaction_id, confidence, real_data_used, result: { risk_score, findings, ... }, provenance }Execution pays automatically from your prepaid credit balance. No human approval is required for normal purchases.
Three calls an agent can make right now
Each example uses a real AOI capability with real data sources. No mock data, no fabricated accuracy. Quotes are free — execution requires prepaid credits.
Assess flood risk for a location
Agent prompt: "Assess flood risk for Zurich, Switzerland"
import { AoiClient } from "@carbonaa/aoi-sdk";
const aoi = new AoiClient({
apiKey: process.env.AOI_API_KEY
});
// 1. Quote (free)
const quote = await aoi.quote(
"flood_risk_score",
{ "location": { "lat": 47.37, "lon": 8.54 } }
);
// → { quote_id, price, currency, valid_until }
// 2. Execute (consumes credits)
const result = await aoi.execute(
quote.quote_id
);
// → { transaction_id, result, provenance }- →result_summary — plain-language risk summary
- →risk_score (0–100) — quantitative flood risk score
- →confidence — model self-reported certainty (NOT validated accuracy)
- →findings[] — specific findings with source references
- →provenance[] — provider, dataset, sensor, observation time
- →data_freshness — "real", "stale", or "model_only"
Check air quality at a location
Agent prompt: "Check current air quality in Paris"
import { AoiClient } from "@carbonaa/aoi-sdk";
const aoi = new AoiClient({
apiKey: process.env.AOI_API_KEY
});
// 1. Quote (free)
const quote = await aoi.quote(
"air_quality_assessment",
{ "location": { "lat": 48.85, "lon": 2.35 } }
);
// → { quote_id, price, currency, valid_until }
// 2. Execute (consumes credits)
const result = await aoi.execute(
quote.quote_id
);
// → { transaction_id, result, provenance }- →result_summary — air quality summary
- →observations[] — PM2.5, PM10, NO2, SO2 from OpenAQ stations
- →provenance[] — provider: OpenAQ, sensor, acquisition time
- →evidence_status — VERIFIED when real station data is used
- →data_freshness — "real" when live station data is returned
- →NO_DATA — reported honestly if no stations nearby
Assess regulatory compliance risk
Agent prompt: "Assess regulatory compliance risk for a steel facility in the EU"
import { AoiClient } from "@carbonaa/aoi-sdk";
const aoi = new AoiClient({
apiKey: process.env.AOI_API_KEY
});
// 1. Quote (free)
const quote = await aoi.quote(
"regulatory_compliance_risk",
{ "facility_type": "steel", "location": { "lat": 51.5, "lon": -0.12 } }
);
// → { quote_id, price, currency, valid_until }
// 2. Execute (consumes credits)
const result = await aoi.execute(
quote.quote_id
);
// → { transaction_id, result, provenance }- →result_summary — compliance risk summary
- →risk_score (0–100) — regulatory exposure score
- →findings[] — specific compliance gaps identified
- →recommendations[] — actionable remediation steps
- →provenance[] — data sources used for the assessment
- →limitations[] — what data was missing or insufficient
Built for autonomous agent commerce
The infrastructure that lets agents transact environmental intelligence safely and automatically.
Autonomous agent execution
Agents execute within their prepaid balance and spending policy. Normal purchases never require human approval.
Deterministic quoting
Every quote returns a binding, exact price before execution. No surprises, no opaque metering.
Idempotent execution
Retry safely. The same idempotency key never charges twice — atomic credit reservation guarantees it.
Atomic credit reservation
Credits are reserved before execution and released on failure. The ledger is always consistent.
Environmental intelligence
38 production capabilities across climate risk, carbon, emissions, vegetation, water, and decision support.
Provenance & evidence
Every result carries confidence, evidence refs, and data provenance. Internal cost and reasoning stay private.
API + SDK + MCP
Three access channels to the same AOI infrastructure. Agent-to-agent by design.
Environmental intelligence for AI agents
AOI is environmental intelligence infrastructure for AI agents, built by Carbonaa SA. It lets AI agents discover, price, and execute environmental intelligence on demand — climate risk analysis, carbon market signals, emissions monitoring, and environmental decision support — through a single quote-first commercial model.
Agents connect through a REST API, an MCP server (Model Context Protocol), or the npm SDK (@carbonaa/aoi-sdk). Every execution returns structured intelligence with confidence scores, evidence references, and data provenance — never raw data dumps, never fabricated results.
The commercial model is quote-first: an agent requests a quote, reviews a deterministic price, then executes. Credits are prepaid and consumed automatically on successful execution. Normal execution never requires human approval. Idempotency keys guarantee no duplicate charges, and atomic credit reservation keeps the ledger consistent.
Explore the capability catalog, read the documentation, see pricing, or connect over MCP.
Enterprise intelligence → SatClimate
AOI is built for developers and AI agents. For enterprise-grade satellite intelligence — scenario foresight, carbon fraud detection, facility monitoring, and market signals — our SatClimate platform provides the full suite with custom integrations and dedicated support.
Explore SatClimateStart building with AOI
Get an API key, install the SDK, and run your first environmental intelligence execution in minutes.
