How to Build an AI Assistant for Instagram DMs (Free Dev Tier, Your Own Logic)

Instagram has 2 billion monthly active users. Roughly 200 million businesses. And the built-in tools for handling business DMs are exactly the same as they were in 2019: you read them yourself and type replies with your thumbs.

There's no away message feature worth using. No way to answer common questions automatically with any intelligence. No booking system built into DMs. Nothing.

Meta gave WhatsApp Business a whole suite of tools (labels, quick replies, away messages, catalogs, template messages). Instagram Business got a "Professional Dashboard" with analytics.

So if you want intelligence on Instagram DMs, you build it. The good news: with the right API layer, that's an afternoon, not a quarter. This post walks through how.

---

What "AI Assistant" Should Mean in 2026

The word "chatbot" makes people cringe, and for good reason. For the last 8 years it meant a rigid script that responds to keywords and breaks the moment someone types a natural sentence.

"Hi, I'd like to book a facial for Saturday"

Old chatbot: "I didn't understand that. Please choose from the options below:

1. Our services
2. Business hours
3. Contact us"

That's not assistance. That's a phone tree without the phone.

In 2026 you can do something completely different, because you control the logic. An incoming DM hits your webhook, you pass the text to the LLM of your choice with your business context, and you send back a real answer. Not menus. Not canned replies. Whatever your code decides.

When someone DMs a salon and says "how much for balayage?", your handler can answer "$180 for medium length, $220 for long. Takes about 3 hours. Want me to check availability?" because you wired it to your pricing data and your LLM. The intelligence is yours. The messaging platform just delivers it.

---

Why Instagram DMs Are Worth Building For

This might sound counterintuitive. WhatsApp gets all the attention in business messaging. But Instagram DMs are a high-leverage place to build, and here's why.

Instagram Has Almost No Built-in Tools

WhatsApp Business gives you away messages, quick replies, labels, template messages, and catalogs. Instagram Business gives you quick replies, and that's about it. There's a real gap here, which means anything intelligent you build stands out immediately.

Instagram Customers Expect Fast Responses

Instagram is a visual, real-time platform. People scroll, discover a business, tap "Message," and expect a response within minutes. They're sending quick, casual DMs:

"Hey, do you have appointments tomorrow?" "How much for lashes?" "Do you do gel nails?"

These are impulse messages. The customer is interested right now. If they don't get a response quickly, they keep scrolling. A webhook-driven assistant that answers in under a second is the whole game.

Instagram Can't Send Proactive Reminders

On WhatsApp, businesses can send template messages proactively. Instagram doesn't support this. You can only respond inside the window after the customer messages you.

This makes the first conversation critical, which is exactly why building intelligence into it pays off. If your assistant handles that first exchange well (answering questions, offering times, confirming a booking right there), you capture the customer while you have their attention. Wait until morning to respond by hand and the window has closed, both literally and figuratively.

---

How to Build It: Connect Instagram, Wire a Webhook

The pattern is the same one you'd use for WhatsApp. Your code doesn't care which channel the message came from. It cares about understanding the question and producing a useful answer.

The Setup (About 10 Minutes)

1. Create a Wabery account. Free developer tier. No credit card.
2. Connect your Instagram Business account. Link through Meta, the same process as any Instagram integration.
3. Point a webhook at your backend. Wabery sends a signed event for every incoming DM through its unified channels API.
4. Ship a handler. Pass the message to your LLM with your business context, send the reply back through the API. Done.

Because Wabery uses one unified channels API, the same handler also covers WhatsApp and Messenger. Connect another channel and your existing code just works.

A Minimal Handler

// Wabery delivers a signed webhook for each incoming message
export async function onMessage(event) {
  // event.channel is "instagram" | "whatsapp" | "messenger"
  const answer = await llm.respond({
    system: businessContext,        // your services, pricing, hours
    message: event.message.text,
  });

  await wabery.send({
    conversationId: event.conversationId,
    text: answer,
  });
}

That's the core. Everything below is just richer logic inside llm.respond and send.

What the Conversation Looks Like

Customer (9:30 PM): "Hey! I saw your reel about the keratin treatment. How much is that and how long does it take?"

Your assistant: "Hi! Our Keratin Treatment is $200 and takes about 2.5 hours. It's one of our most popular services, results last 3-5 months depending on hair type.

Would you like to book one? I can check this week's availability for you."

Customer: "Yes please! Something Thursday or Friday?"

Your assistant: "Let me check. Thursday we have 10:00 AM with Ana and 2:00 PM with Maria. Friday we have 11:00 AM with Ana.

Any of those work for you?"

Customer: "Friday 11 with Ana!"

Your assistant: "You're booked! Here are the details:

Keratin Treatment with Ana Friday at 11:00 AM Duration: ~2.5 hours Price: $200

Quick note, please come with freshly washed hair, no products. I'll remind you Thursday evening. See you Friday!"

That's a $200 booking, captured from an Instagram DM at 9:30 PM, triggered by a reel. The availability check and the booking write hit your calendar, because that's logic you control. Wabery delivered the messages; your code did the thinking.

---

Cross-Channel: Instagram to WhatsApp

Here's a pattern worth building: Instagram has no template messages, so no proactive reminders. WhatsApp does. You can bridge the gap yourself.

When a customer books through Instagram, collect their phone number in the conversation. A clean way to do this is a WhatsApp Flow, Wabery's native in-chat form, but you can also just ask:

Your assistant: "You're all set for Friday at 11 AM! One quick thing, would you like a WhatsApp reminder the day before? If so, what's your phone number?"

Customer: "Sure, it's 555-0123"

Now you store the number against the customer's profile. Your reminder job sends the day-before nudge over WhatsApp, where proactive messaging is fully supported. Discovery and the first conversation happen on Instagram; the ongoing relationship continues on WhatsApp.

This isn't two systems. It's one unified channels API, and you decide which channel to use for what:

• Instagram: Discovery, first conversation, booking
• WhatsApp: Reminders, follow-ups, ongoing communication

---

What About Facebook Messenger?

Same story. Connect your Facebook Page and your existing webhook handler covers Messenger conversations too. Same API, same code, different channel.

Messenger has slightly more outbound flexibility than Instagram (limited proactive messaging through approved tags), but the build is identical: message in, your logic runs, reply out.

Three channels. One API. One handler.

---

The Free Dev Tier and What It Covers

Wabery's free developer tier lets you build and test across Instagram, WhatsApp, and Messenger:

• The full channels API so you can connect a real channel and receive live events
• Signed event webhooks so you can build and test your handler end to end
• Native WhatsApp Flows for in-chat data collection
• CLI and MCP server for scaffolding and local testing
• No expiration, no credit card to start

Usage-based pricing kicks in as your volume grows, so you pay for what you ship, not for seats you don't use. The point is: you can have a working AI assistant on your own Instagram DMs before you spend a cent.

---

Why Build This Now

Instagram's user base keeps growing. DMs are becoming the primary way customers reach businesses they discover on the platform. And Instagram has shown no sign of building meaningful business messaging tools.

That gap between how customers want to communicate (instant DMs) and what Instagram ships (nothing) is only widening. It's a gap you can fill with a weekend of work and own completely.

The businesses that respond to a Tuesday night DM with "Thanks, we'll get back to you during business hours" are handing customers to whoever responds at 10 PM with real answers. If you're the developer who builds that 10 PM responder, you win those customers.

---

Getting Started

Here's the build, start to finish:

1. Sign up at app.wabery.com. Free dev tier.
2. Connect Instagram. Connect WhatsApp too if you want the cross-channel follow-up.
3. Point a webhook at your backend and ship the handler above. 5 to 10 minutes.
4. Your AI assistant is live on your DMs, running your logic.

For a deeper comparison of building your own assistant vs. closed chatbots, read AI Agent vs Chatbot: What's the Difference. For a tour of Instagram DM tooling, see Instagram DM Tools: What Actually Works in 2026.

Your Instagram DMs are already full of potential customers. The question is whether you build something that answers them or leave it to silence.