Instagram is one of the highest-intent discovery channels for e-commerce. Someone sees your product post, wants to know more, and sends a DM. If that DM sits unanswered for hours, the moment passes. This playbook sets up your AI Agent to catch that moment, answer the question, and move the contact toward a purchase automatically.
What you'll build
An AI Agent that acts as a knowledgeable shopping assistant inside your DMs. It answers product questions, helps people find the right item, handles common objections, and sends a checkout link when the contact is ready to buy.
Step 1: Build a product-focused knowledge base
In AI Agent → Knowledge access and behaviour, treat the knowledge base as your product catalogue. For each product or collection, add the name and description, key features, sizing or materials, price and current offers, who it's best for, and honest answers to common questions ("Does it run small?", "How long is shipping?").
If you use Shopify, connect your store first so product data syncs in. See Shopify: Connect Your Store and Send Product Links.
Step 2: Add your checkout links as intent-based actions
In AI Agent → Actions and automations, create an action for purchase intent ("contact is ready to buy" or "asks how to purchase") and attach a Conversion link or Payment link pointing to checkout. If you sell several products, create a separate intent and link for each main product so the agent sends the right one. See Intent-Based Actions: Links, Folders & Scenario Triggers and, for the link types, Conversion Actions: Links, Payments, Bookings & Surveys.
Step 3: Set the personality for selling
In AI Agent → Voice and personality, pick the Sales closer role and keep the tone warm and genuinely enthusiastic about the products, never pushy. Tell it to help people find the right item, answer honestly, and make buying easy when they're ready.
Step 4: Set the goal and the rules
In AI Agent → Goals and instructions, set the goal to: help contacts find the right product and guide them to purchase. In Account description, describe your brand and typical customer. Use Other instructions for the e-commerce edges: what to say when something's out of stock (suggest the closest alternative), how to handle returns (share the policy, promise nothing beyond it), and what never to do (never confirm delivery dates, always point to the shipping policy).
Step 5: Set activation rules
In AI Agent → Activation rules, use Specific intents only with intents like "asks about a product", "asks about pricing", "wants to order", or "asks about shipping", so the agent only activates for shopping conversations.
Step 6: Catch comments on product posts
Build a scenario that fires when someone comments on a product post, sends an instant reply, then uses Hand over to AI Agent to let the assistant take the conversation from there. This is one of the highest-converting patterns in e-commerce: a comment becomes a personalised DM in seconds. See AI Smart Actions in Scenarios for the handover, and Comment-to-DM: Turn Comments into Customers for the funnel.
How it works in practice
Someone comments "price?" on a product post. The scenario replies and DMs them, then hands over to the assistant. The agent answers the price question, asks what they'll use the product for, recommends the right size from your catalogue, handles a quick objection about shipping, and sends the checkout link when they say they're in.
Variations
Retarget the almost-buyers. Set up a "Clicked checkout" folder via an intent-based action and add contacts when the checkout link fires, so you can follow up with anyone who didn't convert.
Greet ad traffic instantly. If you run Instagram ads, fire a scenario on the ad reply that immediately hands over to the shopping assistant, so high-intent contacts get a relevant response at once. See Trigger scenarios from Ads.
Results to expect
An assistant that answers in seconds at any hour catches buying moments your team would miss overnight or mid-launch. Pre-purchase questions get answered before interest cools, and high-intent DMs convert at a meaningfully higher rate than letting them sit.
🐾 Netsuke's Tips
Add "best for" notes to each product. Beyond specs, telling the agent who a product suits lets it make genuine recommendations instead of listing features.
Add a "gift recommendations" entry for people shopping for someone else. It's high-intent and often goes unaddressed.
Keep promises tight. An assistant that never overpromises on delivery or returns saves you support headaches later.
What's next?
Make sure people who clicked checkout but didn't buy don't disappear. Build a follow-up with Re-engagement Campaigns for Cold Leads, or sell structured offers with Selling Courses, Tickets & Bookings via DMs.