Getting Started with Chat
Your introduction to the AI Agents & Chat section. Learn what Agency Hero's AI chat can do, how to access it, and where to go next.
Agency Hero’s chat is more than a place to ask questions — it’s your primary interface for getting work done across your workspaces. The AI agent on the other side knows your clients, your meetings, your open tasks, and the decisions your team has made. You can talk to it like a trusted colleague who has read everything and is ready to help.
This article is your starting point for the AI Agents & Chat section of the docs. It covers the essentials of how chat works and where to go to learn more.
What makes Agency Hero chat different
Most AI chat tools are generic — you have to explain your situation every single time. Agency Hero’s agent is different because it’s grounded in your real workspace data.
Every conversation happens in context. The agent knows:
- Which workspace you’re currently viewing
- Your meeting history, extracted decisions, and open questions
- Active tasks and who owns them
- Documents, memos, and knowledge your team has built up
- What risks and signals have been logged
You don’t need to copy-paste background. Just ask: “What did we decide last week?” or “What’s still open from the Acme kickoff?” — and the agent finds the answer from your actual data.
How to access chat
The chat panel is always visible inside Agency Hero. It lives in the right column of the interface and never unmounts as you navigate — switch workspaces, open tasks, view meetings, and the chat stays right where it is.
Two layout modes
Work Mode (default) puts your workspace content front and center. Chat sits in a column on the right, with thread tabs at the top so you can juggle multiple conversations.
Chat Mode expands chat to the center of the screen and adds a Threads Rail on the left — a scrollable list of all your past conversations, grouped by date and searchable. Use this when you want to focus on an extended AI work session.
Toggle between modes with Cmd+Shift+C (Mac) or Ctrl+Shift+C (Windows/Linux), or use the Work/Chat toggle in the left navigation rail. Your preference is saved across sessions.
Starting a conversation
Type in the composer at the bottom of the chat panel and press Enter to send. Use Shift+Enter for a line break without sending.
To start a fresh conversation, click [+ New Chat] in the Threads Rail (Chat Mode) or the [+] tab button (Work Mode). Each new thread is a blank slate — but the agent always has access to your current workspace context.
What you can do with chat
The agent handles a wide range of work, adapting to what you need:
🔍 Search and look things up
Ask about anything in your workspaces — past meetings, logged decisions, tracked topics, open risks, or documents. The agent searches across all your knowledge sources and synthesizes the answer.
“What were the key decisions from last week’s Acme meetings?”
“Find everything we discussed about the API migration.”
“What open risks are tracked in this workspace?”
✅ Create and manage tasks
Describe what you need in plain language. The agent proposes a task for your review — you can edit any field before approving. Nothing is created without your confirmation.
“Create a high-priority task for Sarah: finalize the pricing deck by Friday.”
“Show me everything blocked this week across all my workspaces.”
“Turn the action items from today’s standup into tasks.”
📅 Meeting intelligence
Prep for upcoming calls or debrief after them. The agent can pull meeting summaries, surface action items, and help you see patterns across your meeting history.
“Prep me for tomorrow’s Acme quarterly review.”
“What were the action items from Monday’s planning session?”
“Summarize what we’ve discussed about pricing across all client meetings.”
✍️ Generate content and memos
Ask the agent to draft client-facing updates, internal memos, status reports, or email copy — grounded in your real workspace context.
“Draft a memo summarizing our decisions about the redesign project.”
“Write a status update for the Acme account based on this week’s meetings.”
🌐 Portfolio and cross-workspace views
Get a bird’s-eye view across all your clients and projects at once.
“What’s my week look like across all my workspaces?”
“Which clients have had no meetings in the last two weeks?”
“Are there any overdue tasks anywhere in my account?”
How agents work (the short version)
When you send a message, the agent doesn’t just run a search — it figures out the best approach and executes it.
Skills define how the agent behaves for different types of tasks. A meeting-prep request loads different tools and context than a task-creation request. Agency Hero routes your message to the right skill automatically. You can also invoke a skill directly by typing @ followed by its name — like @meeting-prep or @research-assistant.
Subagents handle complex or multi-step work. When you ask for a status update across five client workspaces, the agent can spin up five parallel subagents to do the lookups simultaneously and synthesize the results. This happens behind the scenes — you just get a faster, more thorough answer.
Context is always loaded. Before replying, the agent assembles a snapshot of your workspace — open tasks, recent meetings, key topics, and more — so answers are grounded in your current situation, not generic responses.
Approvals protect you. Read-only actions (searching, looking things up) happen instantly. Write actions (creating tasks, logging decisions) always show you an approval card first. You can edit the details before confirming. Nothing is written without your explicit sign-off.
→ For a full explanation of the agent architecture, skills, and subagents, see Understanding AI Agents.
What’s in this section
The AI Agents & Chat section covers everything you need to get the most out of your AI agent — from the basics to advanced workflows.
Published now
| Article | What it covers |
|---|---|
| **Getting Started with Chat** *(this article)* | Overview of chat, how to access it, and what it can do |
| [Understanding AI Agents](/docs/ai-agents-chat/understanding-ai-agents) | How the agent architecture works — skills, subagents, context, and permissions |
| [Creating and Managing Tasks via Chat](/docs/ai-agents-chat/creating-tasks-via-chat) | How to create, update, complete, and search tasks via conversation — including the approval flow and meeting action items |
Coming soon
| Article | What it will cover |
|---|---|
| How Skill Routing Works | How the agent selects the right skill for your request, and how to invoke skills manually |
| Asking Effective Questions | Tips and patterns for getting better, faster answers |
| Searching and Retrieving Information | How the agent searches across topics, intelligence, knowledge, and meeting data |
| Tool Approvals and Permissions | How the approval system works and how workspace admins configure permissions |
| Working with Chat Threads | Managing thread history, resuming conversations, archiving, and search |
| Chat in Different Workspaces | How context changes as you move between workspaces |
| Cross-Workspace Chat | Running queries and tasks across multiple workspaces simultaneously |
| Advanced Agent Capabilities | Background research tasks, subagent coordination, and power-user workflows |
Next steps
Ready to go deeper? Here’s where to head next:
Related articles
More resources to help you go deeper.