Skip to main content
ModuleX is an AI workflow orchestration platform: you build workflows with natural language or a visual canvas, connect the tools you already use, and run them from chat, the API, or an SDK. This page is the index to all of it — every feature, what it does, and the one page that goes deep on it. Use it two ways: skim the cards for an area you care about, or jump to the feature matrix at the bottom for a single-table view.

How to read this map

Each card names a feature, says in one line what it does, and links to the page that explains it in full. Every feature has exactly one canonical home — start at the linked page when you want the details.

Build workflows

The visual canvas and the nine node types that make up a workflow.

AI that builds and acts

The AI Composer writes workflows for you; the Assistant gets work done in chat.

Knowledge and RAG

Connect your documents so AI answers from your own content.

Integrations

Connect 175 services and call their tools from workflows and agents.

Realtime collaboration

Co-edit the canvas live and watch runs stream as they happen.

Billing and credits

Plans, the credit model, and the wallet for managed usage.

Developer API and SDKs

Run everything from code with the REST API and the JavaScript and Python SDKs.

Security and enterprise

Organizations, roles, encryption, and enterprise deployment.

Build workflows

A workflow is an editable graph of steps. You build it on a visual canvas, connect steps with edges, and pass data between them with references. Each step is a node, and each node writes its result back into the run’s state.

Workflow builder

The visual canvas where you lay out, connect, and run workflows.

Canvas and controls

Pan, zoom, select, the detail panel, and keyboard shortcuts.

Variables and references

Pull a prior step’s output into a later one with the reference syntax.

Versioning and history

Workflow versions, deployments, and the canvas edit history.

Error handling and retries

How node failures surface, retry behavior, and debugging failed runs.

Workflows and runs

The core mental model: what a workflow is and what a run is.

The nine node types

A workflow is built from nine node types. Each one does a specific job and writes its result into the run state under its own id.

LLM node

Call a language model with prompts, variables, and structured output.

Agent node

Run an autonomous step that can call tools and loop.

Knowledge node

Retrieve from a knowledge base inside a workflow.

Tool node

Call an integration action from one of the 175 connected services.

Function node

Built-in HTTP requests, webhooks, and schema validation.

Conditional node

Branch on expressions or an LLM decision, or loop over data.

Interrupt node

Pause a run to ask a human a question, then resume (human-in-the-loop).

Transformer node

Reshape, map, and combine data between steps.

Guardrails node

Validate content with JSON, regex, and PII checks.

Run and deploy

Once a workflow is built, you can run it from the builder, expose it as an API endpoint, trigger it from chat, or run it on a schedule.

Running workflows

Run from the builder and watch the live stream of events.

Deploy and versions

Deploy a workflow and manage its versions.

Schedules

Run a workflow automatically on a cron or interval schedule.

Run via API

Trigger a workflow programmatically and stream its result.

Run from chat

Expose a workflow as a chat-triggered run.

AI that builds and acts

ModuleX has two AI helpers. The AI Composer turns plain English into an editable workflow graph. The Assistant is an agentic chat that uses your connected tools to get work done, with no workflow required. Both pause for your approval before sensitive actions.

AI Composer

Describe a process in a prompt and get an editable workflow back.

Composer in the builder

Generate and edit your workflow by chatting with the Composer copilot.

Assistant

An agentic chat that searches tools, drafts outputs, and acts step by step.

How the Assistant works

The agentic loop: how it reasons, picks tools, acts, and decides it is done.

Assistant tools

How the Assistant discovers and calls integration tools.

Human-in-the-loop

How the Assistant pauses to ask for input or approval, and how you respond.

Streaming responses

Stream the Assistant’s actions and output live, in the app and the SDKs.

Models and settings

Choose the model and configure how the Assistant behaves.
The AI Composer and the Assistant are two sides of the same agent. The Composer edits a workflow graph for you; the Assistant has no workflow tools and instead acts directly through your connected integrations. Start with AI Composer and Assistant to learn which fits the job.

Knowledge and RAG

Connect your own documents and ModuleX retrieves the relevant context automatically — in chat, in the Assistant, and inside workflows. You can use ModuleX-managed storage (billed in credits) or bring your own vector store.

Knowledge and RAG

How retrieval works: knowledge bases, ingest, and answering from your content.

Knowledge overview

Manage the knowledge bases that power retrieval across chats and workflows.

Managed knowledge

ModuleX-hosted vector storage and retrieval, billed in credits.

External providers

Bring your own vector store: Qdrant, Pinecone, MongoDB Atlas, or Weaviate.

Managing documents

Upload documents, watch them process, and manage chunks.

Chat with your knowledge

Ask questions answered from your connected knowledge bases.
The document upload limit is set by your plan’s entitlement, not a fixed file size. See managing documents for what your plan allows.

Chat

The chat surface is where you talk to the Assistant, run workflows, and query your knowledge. Chats can be personal or shared with your organization.

Chat overview

The chat surface for the Assistant, workflow runs, and knowledge.

Model selection

Choose the model for a chat — managed models or your own keys.

Run a workflow from chat

Trigger and watch a workflow run directly inside a chat.

Voice messages

Send voice input in chat.

Private and org chat

Personal versus organization-scoped chats and visibility.

History and folders

Organize chats with history and folders.

Integrations

Connect ModuleX to the services you already use, then call their actions from workflows and the Assistant. The catalog has 175 integrations.

Integrations overview

Connect external services and call their tools from workflows and agents.

Integration catalog

Browse all 175 integrations by category.

Authentication and credentials

How integrations authenticate: API keys, OAuth2, and the auth schema variants.

Managing credentials

Create, rotate, and scope credentials in the app and via the API.

LLM providers

Connect language-model providers — managed or bring your own key.

Knowledge providers

The vector stores ModuleX can use for retrieval.

Build an integration

Author your own ModuleX integration and expose its tools.

Custom MCP servers

Connect an external MCP server and use its tools in ModuleX.

ModuleX MCP

Publish your workflows, Files, and Knowledge as tools for external AI clients.
ModuleX does not yet ship per-tool install extras. Installing with extras like [all] or [github,slack] installs only the core package and can emit pip warnings. Install the base package instead — see installing integrations.

Realtime collaboration

Work on a workflow with your team in realtime, and watch runs stream as they happen. The canvas syncs over a live connection so everyone sees the same graph.

Realtime and collaboration model

How runs stream and how the canvas syncs between people.

Canvas collaboration

Multiple people editing the same workflow canvas at once.

Realtime co-editing and external sync

How canvas edits sync between collaborators and from external changes.

Presence, locks and versioning

Presence, cursors, node locks, and the conflict model.

Chat collaboration

Shared and org-visible chats for teams.

Collaboration walkthrough

Invite a teammate and co-edit a workflow live, step by step.

Billing and credits

Managed usage — runs, managed models, and managed knowledge — is priced in credits. Each organization gets a monthly credit allowance from its plan, and paid plans can spend a prepaid wallet beyond it. Bring-your-own-key usage is billed by the provider directly and is not metered in credits.

Billing and credits overview

How ModuleX bills: plans, credits, the wallet, and the usage gate.

Plans and pricing

Free, Pro, Max, and Enterprise — allowances, limits, and pricing.

Credits and metering

What a credit is and exactly what consumes credits.

Usage gating and limits

The admission gate and its denial responses on managed surfaces.

Wallet and top-ups

The prepaid wallet, manual top-ups, and auto-topup.

Subscriptions and Stripe

Subscription lifecycle, Stripe, and the customer portal.
Managed runs, the Assistant, the Composer, and managed knowledge pass through a live billing gate. When credits run out, those surfaces return a denial response (402, 403, or 429). Plain settings and CRUD routes are not gated. See usage gating and the errors reference.

Developer API and SDKs

Everything you do in the app is available from code. Authenticate with Authorization: Bearer mx_live_… plus the X-Organization-ID header, then call the REST API directly or through the JavaScript or Python SDK.

API overview

Base URLs, the request lifecycle, and how every operation is shown.

Authentication

Authenticate every request with a bearer key and the org header.

Errors and status codes

The error-envelope shapes and which surface emits each.

Rate limiting

Per-key and per-user rate limits, and the 429 responses.

SDKs overview

The official JavaScript and Python SDKs, used across every operation.

SDK and API parity

A side-by-side map of REST routes to SDK methods, with gaps called out.

Streaming and HITL

Consume run streams and answer human-in-the-loop prompts from the SDKs.

Realtime overview

The two realtime planes: SSE run streaming and Socket.io collaboration.
A single workflow run, shown three ways. Swap the workflow id and your key:
The subscriptions resource is available in the Python SDK only. Manage subscriptions from Python or call the REST routes directly — see the parity matrix.

Security, organizations, and enterprise

ModuleX is multi-tenant: every workflow, credential, knowledge base, and wallet belongs to an organization. Access is governed by org roles, and credentials are encrypted at rest.

Security overview

How ModuleX secures your data, credentials, and access.

Organizations, roles and membership

Tenancy in ModuleX: organizations, the org header, and roles.

Roles and permissions

Which actions require owner or admin.

Data security and encryption

Encryption of credentials and secrets, and key management.

Trust and compliance

Data handling, retention, and compliance posture.

ModuleX for enterprise

Collaboration, governance, and scale for larger teams.
Organization roles are owner and admin. The Composer, the Assistant, schedules, and knowledge management require owner or admin. The older member role has been retired. See roles and permissions.

Behaviors to know

A few behaviors are worth knowing before you build — see Behaviors & edge cases for the full list.
  • Tool dependencies install directly — install the base package, then add each tool’s SDK.
  • OAuth credentials: if a connection expires, reconnect the integration to issue fresh tokens.
  • The subscriptions resource exists in the Python SDK only, not the JavaScript SDK.
See Behaviors & edge cases for the full list.

The feature matrix at a glance

One table, every area. Use it to find the page that goes deep.

Where to go next

How ModuleX works

The end-to-end mental model behind every feature here.

Recipes

Reusable patterns that combine features to solve real problems.

Optimization

Tune workflows and usage for speed, reliability, and cost.