The AI workspace that's connected to your tools and handles the bills.

Start free Explore integrationsFree plan. No credit card.

175 integrations. 1,758 actions.Browse the catalog

A sentence in. A system out. The Composer is the AI builder inside ModuleX. It turns a description into a plan you can read, then a workflow graph you can edit node by node.

Describe

Write the process the way you would brief a teammate. Name the tools, the conditions, and what done looks like.

“When a Stripe invoice fails, look up the customer in HubSpot, draft a reminder with Claude, send it from Gmail, and post the case to #ar-review.”

to: Composer

Review the plan

The Composer drafts the steps and asks before it changes anything. Open any node, adjust the logic, swap a tool.

Trigger · Stripe invoice.payment_failed

Enrich · HubSpot contact lookup

Draft · Claude reminder emailediting

Send · Gmail + Slack #ar-review

Deploy

One click makes it live. Trigger it from chat, call it from the API, or put it on a schedule.

v3 · production

chat · “run invoice follow-up for Acme”POST /v1/executionsschedule · Mon 09:00

Start building free

composer · acme workspace7 connected

“When a Stripe invoice fails, look up the customer in HubSpot, draft a reminder with Claude, send it from Gmail, and post the case to #ar-review.”

to: Composer

from your workspace

org vault

knowledge

billing-policies.pdf Google Drive

dunning-runbook.md

Composer built · lead-routerv3 deployed

Triggerinvoice.payment_failed

Enrichcontact + plan lookup

Draftreminder · billing-policies.pdf

Emailsend to customer

Notifypost to #ar-review

A real canvas, not a black box. Every workflow is an editable graph. Inspect what the Composer built, rewire a step, and pin any step to your own documents so answers come from your knowledge base, not from model memory.

Open any node to see its exact tool call, inputs, and outputs.
Attach knowledge bases and the workflow retrieves your documents at run time.
Version and deploy; the previous version stays one click away.

Your team runs it. You stay in control. Credentials live in an org vault, so teammates run workflows without ever seeing a key. Every execution lands in an audit ledger with its user, tool, and cost.

Org credential vault: OAuth and API keys stored once, masked everywhere.
Role permissions and approval steps for sensitive actions.
Per-execution ledger: who ran what, with which credential, at what cost.

Credential vaultorg · encrypted

SlackOAuth 2.0org-wide
OpenAIAPI key••••••6H2A
HubSpotOAuth 2.0sales team

Audit ledgerper execution

maya@acme.devran invoice-follow-up12 cr
jon@acme.devedited node · Draft reminder·

RolesAdminEditorRunnerApprover

Managed modelsbilled in credits

ModulexAI21 managed models, billed in credits

in use

Your provider keyspay providers directly

Anthropic
OpenAI
Gemini
xAIconnect
OpenRouterconnect
AWS Bedrockconnect

keys encrypted in the org vault, used without being seen

Your keys or ours. Your data either way. Start on ModuleX-managed models with zero setup, or connect your own provider accounts and pay providers directly with no markup on that usage.

Keys are encrypted in the org vault; teammates use them without seeing them.
Every run is attributed in the audit ledger, per user and per credential.
Role permissions decide who can connect, edit, and run.

Running procurement? Talk to sales

The whole platform, behind an API. Everything the canvas does is callable. Trigger executions from your backend, stream results, and manage workflows with the JavaScript or Python SDK.

Read the docs Get an API key

run.ts

import { Modulex } from 'modulex-js';

const client = new Modulex({ apiKey: process.env.MODULEX_API_KEY });

const run = await client.executions.run({
  workflowId: 'invoice-follow-up',
  input: { customer: 'acme' },
});

console.log(run.status); // "succeeded"

Run it from anywhere.

The same workflow answers a chat message, a POST request, and a Monday 09:00 schedule. Runs stream their steps, costs, and outputs as they happen.