# Actively AI

> Actively AI provides "Intelligence-Led Revenue" — an AI revenue-intelligence
> platform whose Per-Account Agents work every account 24/7, guiding sales teams
> on the next best action and helping them execute it across the prospect and
> customer lifecycle.

This file is GENERATED by the API Evangelist enrichment pipeline from live probes
of Actively's own hosts and the artifacts captured in this repository. Actively
does not publish an /llms.txt of its own (404 on every host as of 2026-08-13).

## Machine-callable surface

Actively ships exactly one machine-callable API surface, and it is an MCP server.

- **MCP endpoint: `https://api.actively.ai/mcp`** — a LIVE hosted (remote) Model
  Context Protocol server named "Actively Intelligence MCP". Transport is
  streamable HTTP; a legacy SSE path exists at `/mcp/sse`. There is no stdio or
  npx package — the server is remote-only.
- **Authentication: OAuth 2.1**, authorization code + PKCE (S256), bearer token in
  the `Authorization` header. Authorization server: `https://auth.actively.ai`.
  Dynamic client registration (RFC 7591) is supported at
  `https://auth.actively.ai/oauth2/register`, so an MCP client can register
  itself with no prior contact with Actively. A human Actively account is still
  required to consent.
- **Tools: UNKNOWN.** `tools/list` returns HTTP 401 (`invalid_token`). The tool
  names and input schemas require an authenticated introspection and are NOT
  listed here, because Actively publishes them nowhere else.
- **Scopes:** only the OpenID Connect quartet is advertised — `openid`, `profile`,
  `email`, `offline_access`. No resource-specific scopes exist.

### Discovery documents (all anonymous, all HTTP 200)

- [Protected resource metadata](https://api.actively.ai/.well-known/oauth-protected-resource) — RFC 9728
- [Protected resource metadata (MCP-scoped)](https://api.actively.ai/.well-known/oauth-protected-resource/mcp)
- [Authorization server metadata](https://api.actively.ai/.well-known/oauth-authorization-server) — RFC 8414
- [OpenID configuration](https://auth.actively.ai/.well-known/openid-configuration)

## Products

- [MCP Server](https://www.actively.ai/products/mcp-server): connects Per-Account
  Agents into ChatGPT, Claude, and Cowork as an intelligence layer for AI agents.
  This is the surface described above.
- [API Platform](https://www.actively.ai/products/api-platform): marketed as
  direct access to per-account agent memory, decisions, and strategy for
  embedding into CRM views, Slack alerts, dashboards and custom applications.
  **No OpenAPI, no base URL, no reference documentation and no SDK is published
  for this product** — access is via a sales/demo conversation.
- [Agent Inbox](https://www.actively.ai/products/agent-inbox)
- [Assistant](https://www.actively.ai/products/assistant)
- [Watchtower](https://www.actively.ai/products/watchtower)

## What is NOT published

An agent or integrator should know these are absent, verified 2026-08-13:

- No OpenAPI / Swagger / GraphQL schema on any host.
- No developer portal, documentation host, API reference, or getting-started guide.
- No SDKs in any package registry (npm, PyPI); the `@actively` npm scope is unregistered.
- No CLI, no Postman collection, no sandbox or test environment.
- No pricing page, no self-serve sign-up, no free tier or trial — contact sales only.
- No status page (`actively.statuspage.io` is an UNCLAIMED subdomain serving
  Atlassian's own marketing page — it is not Actively's status page).
- No changelog, deprecation policy, SLA, or versioning policy.
- No published rate limits, and no rate-limit headers on any observed response.
- No `/.well-known/security.txt` on any host, despite an active disclosure policy.
- No A2A agent card at `/.well-known/agent-card.json` or the legacy `/.well-known/agent.json`.

## Company

- [Website](https://www.actively.ai/)
- [Blog](https://www.actively.ai/blog)
- [Customers](https://www.actively.ai/customers) — Ramp, Samsara, Verkada, Ironclad, Greenhouse, Justworks
- [Careers](https://www.actively.ai/careers)
- [GitHub](https://github.com/actively-ai) — 11 public repos, mostly forks; no client library
- [Request a demo](https://www.actively.ai/demo)
- [Login](https://app.actively.ai/login)

## Security & Compliance

- [Trust Center](https://trust.actively.ai/) — Vanta-hosted, on Actively's own domain
- [Security](https://www.actively.ai/security): SOC 2 Type II (Security category,
  audited annually), ISO 27001 and HITRUST (data-center scope), GDPR (with DPA),
  CCPA. TLS in transit, AES-256 at rest in Google BigQuery.
- [Responsible Disclosure](https://www.actively.ai/responsible-disclosure):
  security@actively.ai; scope *.actively.ai; safe harbor; no bug bounty.
- [Terms of Service](https://www.actively.ai/terms)
- [Privacy Policy](https://www.actively.ai/privacy)

## Notes

- Backed by Bain Capital Ventures. Sector: AI apps / go-to-market intelligence.
- Discovery is open, access is gated: an agent can find and negotiate auth with
  the MCP server anonymously, but consent requires an Actively customer account.