Your AI Agent handles Instagram DMs for you. It reads incoming messages, works out what the person wants, and replies in your voice, with your goals, your knowledge, and your rules. This guide walks you through setting it up for the first time with the 6-step wizard.
Before you begin
Your Instagram account needs to be connected first, or the agent has nothing to reply to. If you haven't done that yet, see How to Connect Your Instagram Account to Inrō.
The AI Agent is included at no extra cost on the Pro plan. On the Free plan, you get 50 AI credits a month to try it out, and a trial gets 550. Each agent reply uses 1 credit. When the month's credits are spent, the agent stops replying until they renew, and its settings don't change. Testing it in the wizard's chat or with Chat with Agent doesn't use credits. See AI Credits: What They Are and How to Check Them.
Open the setup wizard
Click AI Agent in the left sidebar. If you haven't set up your agent yet, you'll land on a welcome page. Click Set up your AI Agent to open the 6-step wizard.
If you use Claude or ChatGPT with Inrō, you can also ask it to set up the agent for you. See Configure Your AI Agent via API & MCP.
The wizard shows a live test chat on the right the whole way through. Type a message there at any step to see how your agent responds as you build it.
Step 1: Voice and personality
This step shapes how your agent speaks. Inrō reads your account and recent messages to suggest a starting personality, which you can adjust.
About: who you are, what you offer and who you talk to.
Message length: choose Short (around 10 words), Normal (around 25 words), or Long (around 50 words), depending on how your brand usually writes.
Auto-like positive messages: turn this on if you want the agent to react with a heart to friendly incoming messages.
Message examples: paste in a few real messages you've sent. These show the agent how you actually write, which matches your style better than any description.
Voice rules: how your agent writes, one rule each, like "Never use exclamation marks".
For more on shaping the voice, see Configuring Your AI Agent's Personality and Tone. Click Continue when you're done.
Step 2: Goal and instructions
Here you tell the agent what it's working toward and the rules it always follows.
Goal: what you most want the agent to achieve, like "book discovery calls" or "turn followers into customers."
Instructions: rules it follows everywhere, one per line, like "never discuss competitor pricing."
Click Continue.
Step 3: Training and knowledge
This is where you give the agent information to answer from. When someone asks about your pricing, returns, or how a product works, the agent searches this knowledge to find the answer.
Click Import new item and choose:
Manual text: paste in content directly, like an FAQ or product description.
Text document: upload a file (.txt, .pdf, .doc, or .docx).
Website: enter a URL and Inrō imports that page's content.
You can skip this and add knowledge later. For the full picture, see Knowledge Base: Teaching Your AI Agent What to Know. Click Continue.
Step 4: Actions and automations
This sets up things the agent can do mid-conversation when it spots a specific intent. Click Add an action and choose:
Send a conversion link: the agent shares one of your conversion links when a contact asks something relevant.
Start an automation scenario: the agent triggers a scenario you've built.
Add to folder: the agent moves the contact into a folder based on what they say.
For each action, describe the intent under "When contact wants to…", then set what happens. You can skip this and add actions later. Click Continue.
Step 5: Activation rules
This is where you decide when the agent replies. The wizard offers the two most common choices:
All pending messages: the agent replies to every message a scenario hasn't already handled.
Specific intents only: the agent replies only when a message matches an intent you define. Leave the intents empty and it won't reply at all, which is handy for testing. If you continue without intents, the wizard warns No intents added and asks you to confirm.
There are two more modes (Automation only and Disable all replies) plus folder rules available from the dashboard. Automation only works through AI steps in scenarios, which are skipped on the Free plan, so on Free that mode leaves the agent silent. For all four and how to combine them, see Activation Rules: Controlling When Your AI Agent Replies. Click Continue.
Step 6: Setup complete
You'll see a summary of your configuration. Review it, click Confirm, and your agent is live.
Your AI Agent dashboard
After setup, clicking AI Agent opens your configuration screen, grouped into Behaviour and personality, Capabilities and actions, and Knowledge. From here you can edit any part of the setup, add intent-based actions, set up handover to a human, and choose whether the agent answers only from your knowledge or also from its general knowledge.
At the top right, the Chat with Agent button lets you test replies anytime without waiting for a real follower.
🐾 Netsuke's Tips
The Message examples field pulls more weight than the voice rules alone. Paste in 3 to 5 real messages from different situations (a complaint, a sales follow-up, a warm welcome) and the agent picks up how you actually write.
Use Chat with Agent right after setup. Ask something a confused follower might ask, and check the answer is accurate and on-brand before you go live.
Picking Specific intents only doesn't limit what the agent knows, only what makes it respond. Once a matching message gets it talking, it can answer anything within its knowledge.
The agent won't reply if a scenario already answered that message. If both are running and the agent seems quiet, that priority is usually why.
What's next?
Go deeper on the parts that matter most: Configuring Your AI Agent's Personality and Tone for voice, or Activation Rules: Controlling When Your AI Agent Replies to fine-tune exactly when it speaks.