# Introducing Behavly Agent > Behavly Agent is the AI built into Behavly: ask in plain words and it does the work in your workspace. It looks things up, builds campaigns and drafts replies, and changes nothing until you confirm. Source: https://behavly.io/blog/introducing-behavly-agent · Behavly blog · Product · November 6, 2026 · By Ardit Karavidaj, Founder & CEO, FoundryX **Behavly Agent** is the AI built into Behavly. Ask in plain words and it does the work in your workspace: it builds a campaign, finds a pattern, drafts a reply, reports how a campaign is performing, or explains how any part of Behavly works. Instead of finding the right screen, you say what you want. Behavly Agent looks it up or does the work in your workspace, with your role's access, and it changes nothing until you confirm. #### Ask in your own words Open **Behavly Agent** from **Home** and type what you need. Answers stream back as they are generated. A few of the things you can ask for: - **Build a campaign.** Ask for a Pattern Campaign that brings quiet users back, and it builds the campaign as a draft. Nothing sends until you activate it. - **Find a pattern.** Ask for the people who keep coming back to your pricing page, and it finds them as a pattern a campaign can start from. - **Draft a reply.** Ask it to answer someone in the Unibox, and it writes the draft for you to check and send. - **Report on results.** Ask how your latest campaign did, and it answers with the figures from your workspace. - **Explain Behavly.** Ask what **Review** does, or how a Trigger Campaign starts the moment your product sends a trigger like a failed payment, and it explains. The home screen also offers ready prompts to start from, such as **Campaign Review** for rates, drop-off and three fixes, and **Find High-Intent Visitors** for repeat pricing visitors, as a pattern. _The Behavly Agent home: ask in your own words, or start from a ready prompt._ #### It answers from your workspace as it is now Behavly Agent reads the same workspace you see in Behavly: campaigns and their results, contacts and leads, patterns, the Unibox, templates, Insights, email accounts and settings. It looks things up at the moment a question needs them, so every figure it gives comes from your workspace as it is now. When a lookup finds nothing, it says so. It works only in the workspace you have open, with your role's access. On a view-only role, lookups work and changes are refused. To work in another workspace, switch from the profile menu and start a new chat. #### Nothing changes until you confirm Asking never changes anything on its own. When a request would create, change, delete or send something, Behavly Agent shows a **confirmation card** with exactly what will happen and which plan allowances it uses. It runs only when you choose **Confirm**. **30 min** A confirmation card waits this long for your answer. Left unanswered, it expires. Every change made this way is recorded in **Settings → Account → Audit Logs**, marked **via Behavly Agent**. Two more rules hold in every conversation: - It never asks for or accepts a password, an API key or a card number. Connecting a mailbox and paying for a plan happen on their own screens, and it gives you the link. - It treats what it reads as information, never as instructions. A reply from a stranger or a contact's name cannot tell it what to do. #### It writes the emails on your paths Behavly Agent is also the writer inside your campaigns. Every campaign is a path: who enters, then the emails, waits and branches each person goes through. A **Behavly Agent** step on that path is an email written for each person, in your sender's voice, from what they did on your site. **Instructions for Behavly Agent** say what the email is for, such as a check-in for someone who stopped at the setup step. Left empty, Behavly decides from what each person did and where they are on the path. A new campaign opens with **Describe the path**: say who it is for and what each email should do, and Behavly lays out the emails, waits and branches for you to check. Before you activate, **Preview** writes any email for a real person without sending it. With **Review** on, every email Behavly writes waits for you, then sends when you release it. #### The same actions in your AI apps The chat is one way to ask. Claude, ChatGPT, Gemini, Perplexity, Grok, Cursor, Claude Code, VS Code and other AI apps that support MCP can connect to Behavly's **MCP server** and use the same actions as Behavly Agent: campaigns, contacts and patterns, the Unibox, templates, insights, email accounts and settings. Behavly offers about 160 tools to an AI app. The steps are the same in every app; only the app's menus differ. 1. Copy the server URL from **Settings → MCP**. 2. In your AI app, add a new MCP server or custom connector, name it **Behavly** and paste the URL. 3. Connect, choose **Continue with Behavly**, then pick the workspace and **Read & write** or **Read only**. 4. Turn **Behavly** on in a conversation and ask for what you need. An app works with your role and your plan's limits. Deleting something or sending email to real people waits for you: the app gets a link, and you check the change in Behavly and approve or decline it within 30 minutes. Changes made this way show in the audit log as **via MCP**. Any other app or script can use the [Behavly API](https://behavly.io/api/v1/docs), which runs the same actions. Every plan can use the API and MCP. #### It follows your plan and your AI settings The chat is unlimited on every plan, Free included. What it does for you counts exactly as it does everywhere in Behavly: an email it writes is a personalized email, and a contact it adds is a tracked contact. When a limit is reached, it names the limit and links to the upgrade. Behavly Agent uses AI, so it follows your workspace's [AI controls](https://behavly.io/help-center/privacy/ai-controls). When AI is turned off in **Settings → Account → AI & Privacy**, Behavly Agent is off too, and AI apps cannot connect at all. Each conversation is recorded in the AI access log on the same page. #### Get started Behavly Agent comes with every plan, Free included. Open **Behavly Agent** from **Home** and ask for what you need, or start from one of the ready prompts on its home screen. To ask from your own AI app, copy the server URL from **Settings → MCP** and follow [Connect AI Apps And Other Tools](https://behavly.io/help-center/agent/connect-ai-apps-mcp). For what it reads and how it makes a change, see [What Behavly Agent Can See And Do](https://behavly.io/help-center/agent/what-behavly-agent-can-see), or read the [Behavly Agent](https://behavly.io/features/agent) page. Start on the free plan: https://behavly.io/auth/login?mode=signup