Skip to main content

User Data Collection: Saving Contact Info Automatically

Set up your AI Agent to detect and save contact details like email, phone, and custom properties straight from conversations.

As your AI Agent talks to contacts, it can pick up information they share and save it to their contact profile: email addresses, phone numbers, or any custom property you've created in Inrō. You set it up once, and it runs automatically in every conversation from then on.

To get there, go to AI Agent in the sidebar, then User data collection.

The User data collection panel in the AI Agent settings.

How it works

When data collection is on for a property, the agent watches the conversation for that information. If a contact shares it, whether you asked or they mentioned it in passing, the agent detects it, saves it to the profile, and carries on naturally. The contact sees nothing change.

The agent is deliberately careful here. It only saves a value when it's confident, it leaves a field alone if you've already filled it in, and detection re-runs as the conversation grows. It uses AI, so it counts against your AI credits while you're on the free pool.

A contact profile showing a value the agent detected and saved.

Adding a property to detect

Click Add a property to detect. A picker shows the contact properties in your account: the built-in Email and Phone fields, plus any custom properties you've created.

Pick the property you want the agent to watch for. It then appears in the list. You can add as many as you need, and the agent watches for all of them at once across every conversation.

The property picker showing built-in and custom properties.

The two AI switches on a property

A custom property can carry two separate AI settings, and they do different jobs:

  • Accessible by AI agent means the agent can read the property and use its value when it replies. If a contact has "Budget: £50k" saved, the agent can reference that without asking again.

  • AI can capture this property means the agent can detect and save the value when the contact shares it. This one switches on when you add the property here, in User data collection.

Both can be active on the same property, so the agent reads existing values and writes new ones as it picks them up. Properties themselves are defined and stored in your CRM: see Contact Profiles & Custom Properties for how to create them and what property types exist.

Built-in properties you can detect

  • Email. The agent picks up an email address shared in conversation and saves it to the contact's email field. Handy for collecting leads without sending a form.

  • Phone. The agent detects phone numbers in common formats and saves them to the phone field.

Email and phone are the only built-in fields the agent detects. Everything else comes from your custom properties.

Custom properties

If you've created custom properties (things like "budget range", "industry", "product interest", or "location"), add them here too. The agent watches for relevant information and saves it to the right field when it finds a match. For properties with a fixed set of options, like a single-choice or multiple-choice field, the agent only saves a value that matches one of your defined options.

Removing a property

To stop detecting a property, open it and remove it from the list. The agent stops watching for it from that point on. Data already saved to contacts isn't affected.

🐾 Netsuke's Tips

  • Add properties one at a time. Start with Email and Phone, confirm they're being captured correctly, then add custom properties one by one. Going slowly makes problems easier to spot.

  • The agent catches information a contact volunteers, not only what it asks for. If someone drops their email mid-conversation, the agent saves it even though it never asked.

  • Pair data collection with an intent-based action: once an email is saved, add the contact to an "Email collected" folder for a clean segment to follow up with.

  • Check your contact profiles after a few days live. If a property isn't being saved when you'd expect, the contact may be phrasing it in a way the agent doesn't recognise. A knowledge entry describing the expected format can help.

What's next?

Now that the agent can collect data, see how it works alongside your flows. Go to AI Smart Actions in Scenarios to combine the agent and your scenarios.

Did this answer your question?