Overview
Scenarios & Campaigns are two of Inrō’s most powerful tools. They let you build multi-step flows using triggers, delays, conditions, messages, tags, folders, and more.
Now, you can bring your AI Agent directly into these flows in four different ways, depending on how much control or flexibility you want.
Whether you need:
a free-form AI reply,
a structured AI-crafted message,
an intelligent Instagram comment reply, or
a full conversation takeover,
Each AI step behaves differently and serves a unique purpose.
This article walks you through all four AI steps available inside Scenarios/Campaigns- how they work, how they behave, and when to use each one.
1. AI Agent Auto DM
What it does
The AI Agent Auto DM step triggers a single AI reply inside your scenario. It uses your:
Persona & goals
Knowledge & documents
Knowledge access settings
You’re telling Inrō:
“When you reach this step, let the AI answer naturally based on what it knows.”
How it behaves
Reads the last user message + recent conversation context
Uses your global AI Agent configuration only
Chooses tone, structure, and phrasing automatically
Sends one reply (not ongoing)
When to use AI Agent Auto DM
You want natural, open-ended replies (FAQs, support, general questions)
You trust your agent to handle most questions in that part of the flow
You don’t need to control the exact phrasing of the reply
Example use cases
After a user replies to a Story DM, let the AI answer any questions about pricing or availability
In a support scenario, route more complex questions to the AI instead of sending a fixed FAQ message
2. AI Agent Message (Scenario-Specific Instructions)
(One AI reply with exact instructions, tone, and purpose you define)
What it does
Like Auto DM, this step sends a single AI reply — but this time you specify exactly what kind of message the AI should craft.
You’re saying:
“Use my persona and knowledge, but follow these specific instructions for this step.”
In this step, you’ll see an instruction box (see screenshot) where you can define:
The purpose of the message (e.g. “summarize the offer and ask 1 qualification question”)
The tone & length (e.g. “short, friendly, 1–2 sentences”)
Any dynamic fields (e.g. use {{first_name}} or tags from the contact)
How it behaves
Reads conversation + your knowledge for context
Uses global AI settings + your step-specific instructions
Pulls in contact properties + survey answers (if you used them)
Sends one controlled message
Precise distinction vs AI Agent Auto DM
AI Agent Auto DM
Uses your global AI configuration only
The agent decides how to answer based on persona + goals
Best for free-form, open-ended replies
AI Agent Message
Uses your global AI configuration + extra instructions just for this step
You control what kind of message the AI must send (structure, purpose, style)
Best for structured flows where each step has a specific objective
When to use AI Agent Message
You want the AI to send a very specific kind of message at a given step
You’re designing a guided flow (like lead qualification, upsell, or booking)
You still want AI “brain power,” but inside clear, step-by-step guardrails
Example use cases
After a user opts in for a call, send an AI-written message that:
Recaps what they want
Shares your main offer in one line
Asks one key qualification question
In a course funnel scenario, send an AI-crafted DM that:
Links the course
Highlights 2–3 benefits taken from your knowledge
Adapts wording to the user’s previous answers
2.1 Using surveys & contact properties with AI Agent Message
AI Agent Message becomes even more powerful when you combine it with data collected through your surveys and scenario questions.
As users move through your flows, you can:
Ask targeted questions about their industry, goals, budget, level, or challenges
Save each answer as a contact property (e.g.
goal,budget_range,experience_level,hair_type)
Then, in the AI Agent Message step, you can tell the AI to use those properties to personalize its reply.
For example:
“Use the contact’s {{goal}} and {{budget_range}} to recommend the best offer in 2 sentences. If the budget is missing, ask a friendly follow-up question.”
Best practices:
Reference contact properties directly in your instructions
Use answers from surveys to adapt tone and recommendations (beginner vs advanced, low vs high budget, etc.).
This way, AI Agent Message doesn’t just “sound” personal, it actually uses the data you’ve captured to tailor every reply.
3. AI Agent Auto Commenter
(AI replies to Instagram comments publicly and intelligently)
What it does
This step lets your AI Agent reply to Instagram comments, using your knowledge and persona.
It works with triggers like:
Every comment under a post
Comments containing specific keywords
How it behaves
Reads the user’s comment, the post caption, and relevant context.
Uses your persona + knowledge to send a relevant reply
Can direct users to DMs (“Check your inbox!”)
Replies publicly, boosting engagement
When to use AI Agent Auto Commenter
You want to boost engagement on posts without replying manually
You run comment-to-DM campaigns and want smart, on-brand public replies
Example use cases
When someone comments “PRICE?”, the AI replies in comments with a short answer and invites them to DM for details
Under educational posts, your AI replies to detailed questions in comments using your course or service knowledge
4. AI Handover (Conversation-Level Takeover)
(AI manages the full conversation until it ends)
What it does
AI Handover transfers complete control of a DM thread to your AI Agent.
Unlike the other three steps, this is ongoing, not a single reply.
You’re telling Inrō:
“From this step onward, let the AI manage every new message in this conversation.”
From that moment on, the AI will keep replying to new messages in that specific conversation, using:
Your persona & goals
Your knowledge & documents
Your knowledge access settings (strict vs always-answer mode)
This lets you give the AI “ownership” of certain conversations, without turning it on globally for all DMs.
How it behaves
Activates when the scenario hits the AI Handover step
Applies only to that specific DM thread (not your entire inbox)
Uses the same configuration as your main AI Agent setup
Continues responding as long as the conversation is active, until:
The conversation naturally ends, or
Your human-intervention rules/team take over
When to use AI Handover
You want the AI to manage an entire conversation after an initial flow or qualification
You don’t want AI globally on all DMs, but you want full AI handling for certain high-intent threads
You’ve already collected enough information (via questions, surveys, or forms) and want the AI to handle follow-ups, objections, and extra questions
Example use cases
After a lead completes a qualification flow (“QUOTE” scenario), you trigger AI Handover so the agent can:
Answer detailed follow-up questions
Clarify the offer
Keep the conversation warm until they’re ready to book or buy
After a user signs up for a free challenge or mini-course, you use AI Handover so the AI can:
Check in on their progress
Share tips pulled from your knowledge
Guide them towards your main paid offer
AI Handover is a perfect middle ground: you keep your account mostly human-led, but let the AI fully take over specific conversations where it can genuinely save you time.
When to Use Automation-Specific AI Activation
Instead of keeping the AI Agent on for all DMs, you can activate it only inside Scenarios. This gives you granular control over when the AI replies, what it says, and how long it stays active.
Best for
Comment/keyword-triggered flows
Launch funnels
Waitlists & lead magnets
Lead qualification
Complex conversations needing AI takeover
Example Flow: Lead Qualification
Trigger: User DMs “QUOTE”
Ask 2–3 qualification questions
AI Agent Message → summarize + recommend offer
AI Handover → AI manages full conversation until booking
Feedback
If you have questions about any of the AI steps or need help structuring your flows, message us in Intercom. We’re here to help you build automations that feel smart, controlled, and perfectly on-brand.