Developer platform · Nico 2.5 API

Build on Nico 2.5. The GTM model, as an API.

One keyless endpoint runs the full agentic loop server-side against your connectors — sourcing, enrichment, scoring, and drafting — and streams every step back. No model keys to manage, no tool plumbing to build.

Get an API key →Read the docs

Passwordless sign-in · usage metered per token · curl · Python · TypeScript · Go

POST /v1/ask

curl https://nico.macrodeep.com/v1/ask \
  -H "Authorization: Bearer $NICO_SESSION" \
  -d '{
    "model": "nico-2.5",
    "prompt": "Which open deals went quiet this
               month? Draft re-engagement.",
    "stream": true
  }'

# Nico plans, queries your CRM, drafts —
# and holds every send for approval.
event: step        research · 48 open deals scanned
event: step        14 quiet > 21 days · drafts ready
event: approval    14 held — approve in any app

01Choose how you build

Two endpoints. One account & budget.

/v1/askAgentic · managed

Agent API

Hand Nico a goal. It plans the run, calls your connectors in a server-side tool loop, and streams each step — holding every outbound send for approval.

Server-side tool loop over 75+ connectors
Streamed step, tool, and approval events
Approval boundary enforced server-side
Full run trace, replayable and interruptible

Agent API reference →

/v1/chatMetered · you drive

Chat API

Direct, streamed access to Nico 2.5 for classification, scoring, and generation — when you want to own the loop and orchestrate tools yourself.

Streamed tokens with usage per response
Structured output & function-calling schemas
Call the model by id: nico-2.5
Drop-in familiar request shape

Chat API reference →

02From key to production

01 · Get started

Get a key

QuickstartAuthentication

02 · Build

Make a call

Fire /v1/ask or /v1/chat and stream the response. SDKs for four languages.

Agent APIStreaming

03 · Evaluate

Test & measure

Replay runs, inspect tool traces, and score against the GTM eval suite before you ship.

Run tracesEvals

04 · Operate

Ship & monitor

Set budgets, per-model caps, and rate limits. Watch usage and revoke sessions any time.

Limits & budgetsWebhooks

03Platform capabilities

Keyless by design

The platform holds the model keys. You send a session token; we meter every call.

Connectors in the loop

75+ tools callable server-side. Credentials encrypted at rest, never shown to the model.

Approval boundary

Sends and writes are held as approval events — resolve them in any app or via API.

Streaming

Server-sent events for tokens, steps, tool calls, and approvals — render progress live.

Structured output

Constrain responses to a JSON schema for scoring, routing, and enrichment pipelines.

Plan-aware limits

Monthly budgets, per-model caps, and burst rate limits enforced on every request.

Webhooks

Subscribe to run completion and approval events to drive your own workflows.

Revocable sessions

List active sessions, scope them, and revoke instantly. Tokens live in the Keychain.

04Models

One model. One id.

ModelBest forInputOutput

Nico 2.5nico-2.5

The GTM model. Long-horizon agentic runs and the deepest GTM judgment.

$6 / M

$30 / M

Illustrative list pricing per million tokens · tool calls metered separately.Explore the model →

05Keep building

Reference

Ship your first run today.

Get an API key →Explore the model