Can an AI agent call ResearchAO directly?

Yes. Agents authenticate with a scoped personal API key (Bearer rao_live_...) and call documented HTTP endpoints. See /docs/api and /openapi.json.

How do machine agents register?

Machine agents self-register through the public registration flow, complete human-authorised Stripe billing, then receive a scoped credential. Humans sign up normally; machines never get dashboard passwords. Start at /agents/self-register.

Does ResearchAO support scheduled research for agents?

Yes. Create once / daily / weekly / monthly schedules via the API or app so agents can consume completed reports instead of re-running every prompt.

How are credits and rate limits handled?

API research runs consume account credits by effort level. Keys can use scoped permissions and credit limits. Registration and API traffic are rate-limited; treat 429 responses as backpressure.

Do you support MCP or a specific agent framework?

This page does not claim Model Context Protocol (MCP) support or official plugins for any third-party agent framework. Integrate with standard HTTP and OpenAPI.

Developers

Deep research infrastructure for AI agents

ResearchAO gives agents structured, source-backed research over a scoped HTTP API — the same engine as the web product, with schedules, credits, and human-authorised billing.

Connect an agent API docs

Opening answer

AI agents often need multi-source investigation with citations — not another chat completion. ResearchAO exposes deep research as infrastructure: authenticate with a scoped personal API key, start a research run, retrieve a structured report, optionally schedule recurring jobs, and meter usage with the same credits as the web app.

Give an agent a research capability

Agents can call the same research engine humans use in the web app. Practical patterns:

Competitor agent

Instruction example: “Research our three closest competitors every Monday and surface meaningful changes.” Implement with a weekly schedule or API schedule plus a summary step in the agent. Product page: automated competitor research.

Market intelligence agent

“Produce an updated market landscape before our monthly strategy meeting.” See market research.

Due diligence agent

“Research this company using public sources and return areas requiring further verification.” See due diligence research. Do not treat agent output as legal, financial, or investment advice.

Programmatic flow

Agent authenticates with a scoped ResearchAO API key
Agent starts a research job (POST /v1/research)
ResearchAO returns a source-backed report when ready
Agent consumes structured output or a summary

Contracts: /openapi.json, /docs/api, machine registration at /agents/self-register.

What ResearchAO gives an agent

Source-backed deep research reports over HTTP
Scoped personal API keys with permission controls
Run submit, list, retrieve, and follow-up message endpoints
Schedules (once / daily / weekly / monthly)
Webhook delivery on completion (optional)
Credit balance / usage reads
Machine self-registration under human-authorised billing

Auth

Send Authorization: Bearer rao_live_... on API requests. Create and rotate keys in the app. Examples always use placeholders — never commit real keys. Machine agents receive a scoped credential after registration and checkout; they do not use dashboard passwords.

Start research

POST /v1/research with a query and effort level. Persist the returned run id. Prefer an Idempotency-Key header so retries do not create duplicate runs.

Retrieve report

GET /v1/research/{run_id} returns run status and the report when ready. Alternatively configure a webhook so the agent is notified on completion instead of polling.

Scheduling

Use POST /v1/schedules (and related schedule endpoints) for recurring research. Let the agent summarise deltas rather than re-issuing the same prompt every turn. Product overview: /scheduled-research.

Idempotency

Pass Idempotency-Key on mutating calls when your agent may retry. ResearchAO stores idempotency keys so safe retries return the original outcome instead of spending credits twice.

Credits and limits

Research runs consume credits based on effort. Keys can be scoped and credit-limited per agent. Read balances with GET /v1/usage. Choose a plan on /pricing that matches expected agent volume.

Rate limiting

API and registration traffic are rate-limited. Agents should respect limit responses, back off, and avoid tight retry loops that waste quota.

Machine registration

Human owners use browser sign-up. Machine agents use the public self-registration API, then a human-authorised Stripe subscription, before receiving credentials. See /agents, /agents/self-register, and /agents/connect.

Security

Treat API keys as secrets, rotate on exposure, and keep agents on least-privilege scopes. Private app routes stay authenticated and noindex. Overview: /security.

Human-authorised billing

Machine agents do not self-bill without a human. Subscription and portal flows require human-authorised Stripe checkout so spend stays under an accountable owner.

OpenAPI

Contract-first docs live at /docs/api and /openapi.json. Commercial API page: /deep-research-api.

cURL

curl -X POST https://api.researchao.com/v1/research \
  -H "Authorization: Bearer rao_live_..." \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: agent-run-001" \
  -d '{
    "query": "Summarize recent EU AI Act enforcement themes",
    "effort": "standard"
  }'

TypeScript

const res = await fetch("https://api.researchao.com/v1/research", {
  method: "POST",
  headers: {
    Authorization: "Bearer rao_live_...",
    "Content-Type": "application/json",
    "Idempotency-Key": "agent-run-001",
  },
  body: JSON.stringify({
    query: "Summarize recent EU AI Act enforcement themes",
    effort: "standard",
  }),
});

const run = await res.json();
// later: GET /v1/research/{run_id}

Python

import requests

res = requests.post(
    "https://api.researchao.com/v1/research",
    headers={
        "Authorization": "Bearer rao_live_...",
        "Idempotency-Key": "agent-run-001",
    },
    json={
        "query": "Summarize recent EU AI Act enforcement themes",
        "effort": "standard",
    },
)
run = res.json()

Example workflow

Human creates an account (or agent completes self-register + checkout)
Create a scoped API key (or receive the registration credential)
Agent POST /v1/research with idempotency key
Store run_id; poll GET or wait for webhook
Inject the structured report (or a summary) into agent context
For stable questions, create a schedule and only process deltas

Longer guide: Deep Research API for AI agents.

Frequently asked questions

Can an AI agent call ResearchAO directly?: Yes. Agents authenticate with a scoped personal API key (Bearer rao_live_...) and call documented HTTP endpoints. See /docs/api and /openapi.json.
How do machine agents register?: Machine agents self-register through the public registration flow, complete human-authorised Stripe billing, then receive a scoped credential. Humans sign up normally; machines never get dashboard passwords. Start at /agents/self-register.
Does ResearchAO support scheduled research for agents?: Yes. Create once / daily / weekly / monthly schedules via the API or app so agents can consume completed reports instead of re-running every prompt.
How are credits and rate limits handled?: API research runs consume account credits by effort level. Keys can use scoped permissions and credit limits. Registration and API traffic are rate-limited; treat 429 responses as backpressure.
Do you support MCP or a specific agent framework?: This page does not claim Model Context Protocol (MCP) support or official plugins for any third-party agent framework. Integrate with standard HTTP and OpenAPI.

Wire research into your agent

Start from /agents, self-register if you are a machine agent, or open the OpenAPI-backed docs.

Agents overview OpenAPI