Spaces
Use case / Product recommendations

Product recommendations in chat with your AI agent

With Spaces, your AI agent recommends products in the chats customers start on WhatsApp, Instagram, LinkedIn and Facebook Messenger. It asks a few short questions, such as what it is for, which size and what budget, then suggests one or two products that fit, with a photo, the price and a link to buy.

Example

See it in a conversation.

Names and details are examples.

CustomerProduct recommendationsAI agent
Setup

How to set it up step by step.

The full reference is in the partner docs.

  1. 1

    Connect your accounts

    In your Spaces dashboard, connect the accounts you already use.

  2. 2

    Add Spaces to your AI agent

    Add Spaces to the AI agent you already use, such as Claude, ChatGPT or your own code, through the MCP server or the API.

  3. 3

    Give the agent your buying guide

    List the questions that decide the right product, the products that fit each answer, and the photos and links to send.

Details

The details.

Quick answer

Spaces handles the messages. Your catalogue, stock and checkout stay in your shop. The agent learns your range from the notes you give it, or reads it live if you connect your shop to the agent through its own tools or MCP server.

What the AI agent does on each channel

Spaces works with Instagram, WhatsApp, LinkedIn, Facebook and TikTok (coming soon). On Facebook, that means Messenger on a personal Facebook account.

  • WhatsApp: Customers ask “which one should I get?” on the number they already have saved. The agent asks its questions, sends two photos side by side and the links, and tags the chat with what the customer wanted.
  • Instagram: The agent answers direct messages and story replies about a product, and moves a “which size?” comment into a private reply where Instagram allows it, so the advice happens in the messages.
  • LinkedIn: For business buyers, the agent asks about team size, use and budget, then recommends the right plan, model or package and attaches the spec sheet.
  • Facebook Messenger: The agent answers product questions that reach Messenger on your personal Facebook account, with the same short questions and a link to the item that fits.
How can an AI agent recommend products in WhatsApp or Instagram chats?

Write down how a good shop assistant decides, in plain words, and give it to the agent. For shoes, that might be road or trail, distance and budget. For skincare, skin type, the main concern and any allergies. For furniture, room size, style and delivery date. The agent asks those questions one at a time, in the customer’s own chat, and suggests no more than two options, so the customer is not buried in choices.

The answer comes with what makes it easy to buy: a photo, the price and a link. On WhatsApp and Messenger the agent can send photos, voice notes and files in the same chat. On Instagram it can reply to a story or a direct message. It can also save each answer to the customer’s contact, so a later chat starts where this one ended.

What should the agent know about my products?

Start with your best sellers. For each one, the agent needs who it is for, who it is not for, the sizes, the price and the link. Add the questions customers often ask, such as “does it run small?” or “is it safe for sensitive skin?”, and the honest answers. Where you are not sure, tell the agent to say so and pass the chat to a person.

If your range changes often, connect your shop to the agent through its own tools, and it can check what is in stock before it recommends anything. Spaces does not connect to shops, so the agent reads the catalogue itself and only uses Spaces to talk to the customer.

After setup
  • Connect your shop, if you want. To check stock or prices live, connect your shop to the agent through its own tools or MCP server. Spaces does not connect to shops.
  • Test it. Message your own account from another phone or profile as a customer who is unsure, and check the questions and the suggestions in drafts before replies go out.
What always goes to a person

Tell your agent to hand these to a person every time:

  • questions about health, allergies, medical conditions or safety
  • custom orders, made-to-measure work and anything outside the range
  • discounts and price matching
  • a customer who is still unsure after two suggestions
  • complaints about something they already bought
Safe by default, on every account

Recommendations happen in conversations customers start, which is the safest kind of messaging. Replies to someone who wrote in are not held back by the warm-up, so people get advice from the first day.

Spaces paces every message and keeps each account inside its daily limits. Spaces lowers the risk with limits, pacing and warm-up, and pauses sending if an account shows signs of a restriction. No tool can promise that a platform will never restrict an account, and Spaces does not either.

  • A reply to someone who wrote in goes out from the first day and does not count toward the messages a day. Replies have their own generous caps instead (80 a day on days 1 to 3, up to 400 a day), so a long backlog goes out over a few hours rather than in a burst.
  • Instagram replies to people who wrote in: 250 a day and 40 an hour, at least 15 seconds apart (60 a day and 15 an hour on days 1 to 3).
  • Facebook Messenger: up to 50 messages and 8 new chats a day, at least 20 seconds apart.

Every channel’s limits are on sending limits, and how Spaces protects accounts is on account safety.

Read more
FAQ

Questions, answered.

01Can an AI agent recommend products on WhatsApp?

Yes. With Spaces, your agent asks a few questions in the chat and recommends products from your range, with photos, prices and links.

02Does the agent need my product catalogue?

It needs notes on your products, or a connection to your shop through its own tools. Spaces does not connect to shops. The agent reads the catalogue and uses Spaces only for the messages.

03Can the agent send product photos in the chat?

Yes. It can send photos, voice notes and files in the same chat on WhatsApp and Messenger, and photos in Instagram direct messages.

04Will it push customers to the most expensive option?

Only if you tell it to. Give it the rules a good assistant follows, such as staying inside the customer’s budget, and it follows them.

05Can it remember what a customer liked?

Yes. The agent can save answers such as size and preferences to the customer’s contact, with tags and notes, so the next chat starts from there.

06Can I check the suggestions before they go out?

Ask your AI agent to hold replies as drafts, and nothing goes out until you approve it.

07What does it cost?

Free to start, then $19 per connected account a month. Start on the Free plan with no card, and see the pricing page for the details.

Built for

Industries that use this most.

More use cases

Other ways to use Spaces.

Start building.

Start with a workspace of your own to try the API, then email hello@spacesos.com to become a partner.