Docs
Learn how Alloy works
Start with the core ideas behind Alloy: AI teammates, Ally, shared knowledge, approved tools, workflows, webchat, voice, and API access.
Where to start
These docs are a public orientation hub for buyers, users, search engines, and LLMs. They explain the product model first and link deeper for setup details. Pick a section below.
Getting started
- What Alloy isUnderstand the workspace model behind Alloy: people, AI teammates, shared knowledge, tools, and workflows.
- Meet AllyLearn how Alloy's built-in assistant helps users work with knowledge, tools, and AI teammates.
- Add your first AI teammateCreate a focused AI teammate and give it the right role, knowledge, tools, and access.
- Upload and share knowledgeUse Storage to organize files and give AI teammates access to the knowledge they need.
- Invite team membersInvite people to your Alloy workspace and track pending, accepted, cancelled, and expired invites.
Core concepts
- AI teammatesLearn how Alloy uses configurable AI teammates for specific jobs across the workspace.
- AllyLearn how Alloy's built-in assistant helps users work inside the workspace.
- Knowledge and storageLearn how Alloy stores company knowledge and controls what AI teammates can access.
- Tools and MCP serversLearn how Alloy connects AI teammates to approved actions and external systems.
- Skills and workflowsLearn how Alloy turns repeatable work into structured processes.
- Contacts and conversationsLearn how Alloy keeps chat, customer identity, and conversation history connected.
- Permissions and controlLearn how Alloy scopes access for people, AI teammates, storage, tools, and connected systems.
Build and connect
- Webchat widgetEmbed Alloy chat on your website so customers and visitors can talk to your team and AI teammates.
- API accessUse Alloy APIs to manage storage, trigger workflows, and connect external systems to AI teammates.
- MCP serversConnect external tools through MCP and assign them only to the AI teammates that need them.
- Workflow builderCreate repeatable processes that combine AI teammates, tools, logic, messages, and human input.
- VoiceEnable realtime voice conversations with Ally, AI teammates, and customer-facing webchat.
- Bring your own model keysUse your own LLM provider relationship or let Alloy pass model and compute usage through.
Technical docs
- Runtime Tooling For AI TeammatesPublic reference for Runtime Tooling For AI Teammates.
- Alloy AI Node (Agent Step)Public reference for Alloy AI Node (Agent Step).
- Ally, the Built-In Platform AssistantPublic reference for Ally, the Built-In Platform Assistant.
- Auth and SessionPublic reference for Auth and Session.
- Chat WidgetPublic reference for Chat Widget.
- Contacts and ChannelsPublic reference for Contacts and Channels.
- CoworkPublic reference for Cowork.
- Internal ChatPublic reference for Internal Chat.
- Invite Acceptance FlowPublic reference for Invite Acceptance Flow.
- LogsPublic reference for Logs.
- MCP ServersPublic reference for MCP Servers.
- Message ReactionsPublic reference for Message Reactions.
- OmniPublic reference for Omni.
- Private APIPublic reference for Private API.
- Realtime VoicePublic reference for Realtime Voice.
- SkillsPublic reference for Skills.
- StaffPublic reference for Staff.
- StoragePublic reference for Storage.
- Template: Customer Support TeammatePublic reference for Template: Customer Support Teammate.
- Variables & SecretsPublic reference for Variables & Secrets.
- Workflow Engine (Backend)Public reference for Workflow Engine (Backend).
- WorkflowsPublic reference for Workflows.
Frequently asked questions
Who are these docs for?+
Two audiences. The Getting started and Core concepts sections are business-readable and explain the product model in plain language. The Technical docs section is written primarily to guide AI agents — so they can answer questions about Alloy accurately and operate it on the user's behalf. Engineers can read them too; the format is denser than a tutorial because machines do better with terse, well-scoped reference material.
Why is the Technical docs section so terse?+
It's tuned for machine consumption. Short sections, explicit terminology, one capability per page, minimal narrative. AI agents work better with reference material that has clear boundaries — and humans skim it the same way once they know what they're looking for.
What should a new user read first?+
Start with What Alloy is, then read Meet Ally, AI teammates, Knowledge and storage, Tools and MCP servers, and Skills and workflows. By that point you'll understand the workspace model. Drop into the Technical docs section when you need exact operating details for a specific capability.