ARCNM Developer Docs

CAD-to-cost and feature extraction as an API. Three layers work together on every request:

Deep Feature Extraction recognises geometry, PMI, GD&T, and manufacturing intent from your CAD + drawing.
Deep Cost Modeling decomposes setup, cycle time, material, tooling, and overhead.
Adaptive Calibration continuously refines both from your tenant's observed prices — every quote you accept teaches the model.

The result is a defensible, auditable unit price with provenance back to the source of every coefficient.

ARCNM exposes the same engine three ways:

REST — https://api.arcnm.io/api/v1/*
MCP — https://api.arcnm.io/mcp-server/mcp (bearer-token, agent-ready)
Webhooks — push events for wallet and account activity

Billed on success: failed runs, 4xx/5xx, and idempotent replays cost nothing. Read the pricing model →

How ARCNM works

Every quote runs the same pipeline. Call the whole thing in one request with upload-and-quote.

   3D (STEP/STL)   2D (PDF/DXF)         batch size n
        │              │                     │
        ▼              ▼                     │
   ┌──────────────────────────┐              │
   │   Extraction             │              │
   │   features, PMI, GD&T    │              │
   └────────────┬─────────────┘              │
                ▼                            │
   ┌──────────────────────────┐              │
   │   Discipline + Planning  │              │
   │   ops, setups, tools     │              │
   └────────────┬─────────────┘              │
                ▼                            │
   ┌──────────────────────────┐              │
   │   Physics                │              │
   │   times, feeds, wear     │              │
   └────────────┬─────────────┘              │
                ▼                            │
   ┌──────────────────────────┐              │
   │   Calibration            │◀── tenant rates, machines, observations
   └────────────┬─────────────┘              │
                ▼                            │
   ┌──────────────────────────┐              │
   │   Cost                   │◀─────────────┘
   └──────────────────────────┘

Read the concept guides:

Most-used endpoints

Endpoint	What it does	Reference
`POST /api/v1/parts/calculations/upload-and-quote`	CAD → quote in one multipart call	Calculations
`POST /api/v1/parts/calculations/quote`	Quote an existing revision (JSON)	Calculations
`GET /api/v1/parts/calculations/{id}`	Fetch result + `analytics` blob	Calculations
`POST /api/v1/parts`	Register a Part + initial Revision	Parts

Subscribe to event notifications with webhooks. Explore the API interactively in the API explorer, or pull the full machine-readable spec from https://api.arcnm.io/api/v1/openapi-public.json (OpenAPI 3.1 JSON).

Built-in safety nets

Idempotency — Idempotency-Key dedup with a 24h window. Retries are safe.
Errors — every response is { "error": { "code", "message", "details", "doc_url" }, "request_id" }. Stable codes you can switch on.
Rate limits — per-key and per-caller limits; back off on 429.
Versioning — X-API-Version: 2026-04-24 stamped on every response.

SDKs

Native SDKs are on the roadmap. Today, call the REST API directly (it's small) or generate a client from the public OpenAPI spec — see SDKs.

Stay in touch

Status — https://status.arcnm.io
Changelog — the developer changelog
Support — [email protected]
Security — [email protected]

Need a private deployment or a dedicated machine profile? [email protected].