Welcome to EnClaw

EnClaw gives you a team of AI assistants — we call them agents — that you can chat with the same way you'd message a colleague. You ask in plain English; the agent does the legwork: checking your calendar, drafting an email, searching your files, logging a task in JIRA, pulling a report together. You stay in control the whole time — anything consequential waits for your approval before it happens.

You don't need to learn commands or write code. If you can send a Slack message, you can use EnClaw.

If you ever get stuck and this guide doesn't help, your workspace admin or EnClaw support can.

The core ideas

A handful of concepts explain almost everything you'll see in EnClaw.

Agents

An agent is an AI assistant with a defined job — a Project Manager agent, a CRM Assistant, a Customer Support agent. Each one has an identity: a plain-language description of its role, its tone, and its rules. Agents are set up once (by you or your admin) and then you simply chat with them.

The Orchestrator

You don't have to know which agent handles what. When you send a question without naming an agent, the Orchestrator reads it and routes it to the best specialist for the job — including your own personal agents. Think of it as a helpful receptionist who knows everyone in the building. Once it has routed you to an agent, that agent tends to keep answering you in that conversation (see sticky routing) so you're not re-introduced every message.

Personal agents vs org agents

• Org agents are shared with the whole organisation and managed by admins — everyone can use them.
• Personal agents belong to you alone. You create them, you shape them, and they only ever answer you. More in My Agents.

Skills and tools

Skills (sometimes called tools) are the things an agent can actually do — read your Outlook calendar, search the web, create a JIRA issue, send a Slack message. Each agent has a specific set switched on, which is why one agent can book meetings while another can't. Anything that reaches your accounts (your email, your calendar) works through connectors you approve yourself — see My Connectors.

Memory

Agents remember. They keep useful facts, your preferences, and your corrections between conversations, so you don't repeat yourself. Your personal memories stay private to you, and each Slack or Teams channel keeps its own separate context. Details in Memory & context.

Approvals

Some actions — sending an email, creating a JIRA issue, posting to a channel — pause and ask a human to sign off first. That's deliberate: it's how EnClaw keeps agents fast and safe. See Approvals.

How a conversation actually flows

Here's a typical exchange from start to finish:

1. You ask. In Slack, Teams, Telegram or the web app: "What's on my calendar tomorrow, and draft a reply to Priya's email about the workshop."
2. The Orchestrator routes it. Your question lands with the agent best equipped to handle it — perhaps your personal assistant agent, since it involves your own calendar and email.
3. The agent works, and shows its progress. In Slack you'll see a single status message that updates in place as the agent reads your calendar and finds the email. You can type stop at any point to cancel.
4. Anything risky pauses for you. The reply to Priya is drafted, but sending it triggers an approval. You get the draft, you say yes (or no, or "change the second paragraph").
5. You get the result. Calendar summary, email sent, done. The agent quietly notes anything worth remembering — like the fact you prefer morning meetings — for next time.

That's the whole rhythm: ask, watch it work, approve what matters, move on.

What's in this guide

New to EnClaw? Start with the Overview — six short chapters written to be read in order, like a small book, that introduce the ideas (agents, the Orchestrator, skills, memory, guardrails, and where your work lands) before you meet any buttons.

Then the next seven pages mirror the EnClaw sidebar, in the order you'll see it in the app:

In the app	What it covers
Chat	Direct conversations with the model — projects, project context, your conversation list
Dashboard	The landing overview — what the tiles and feeds mean, and which cards are admin territory
My Agents	Your personal agents — templates, cloning, identity, and why they answer only you
My Sessions	Every run you've started, with its files — the "where did my work land" page
Artifacts	Live micro-sites — dashboards, articles, slideshows — created by you or agents
Marketplace	Browsing and installing skills, agents and templates
File Space	Uploads, folders, share links, and where agent outputs live

Then the supporting topics:

Topic	What it covers
Chatting from Slack, Teams & Telegram	Mentions, bound channels, sticky routing, live progress, stopping a run
My Connectors	Linking your Outlook, Gmail, JIRA and more — safely
Memory & context	What agents remember, correcting them, asking them to forget
Approvals	Why actions pause, where requests arrive, what timeouts do
Schedules & reminders	Recurring tasks, one-off reminders, proactive check-ins
Playbooks	Ready-made recipes for sales, marketing, operations, projects and account management
Troubleshooting & FAQ	Quick fixes for the most common hiccups

A few reassurances before you start

• You can’t easily break anything. Reading data is generally free and instant; changing or sending anything important waits for approval.
• Your data stays yours. Agents acting through your connectors see only what you can see, and your personal memories are visible to no one else.
• You can always say stop. Type stop or cancel mid-task and the agent halts.
• Plain words work best. You don’t need magic phrasing. Brief the agent the way you’d brief a capable colleague — context, goal, deadline — and it’ll take it from there.

Ready? Start with Chat — or, if you live in Slack or Teams, jump straight to Chatting from Slack, Teams & Telegram.