By default, every conversation starts fresh. Contact memory changes that: durable notes about a person that AI can read the next time they write in, so a customer doesn't have to explain their setup twice.
It's off until you enable it, under Contact memory in your AI settings. The description there is exact: it only applies to contacts identified via Gleap.identify.
That restriction is deliberate. A guest is a browser session, not a person — remembering things about one would mean carrying them over to whoever uses that browser next.
A Memory tab then appears on identified contacts.
Each entry has a name, a one-line description of what it's about, and the content the AI actually reads. Think “what setup are they running”, “what they've already tried”, “what they told us about their rollout” — facts that stay true after the conversation ends.
Every entry shows who wrote it: your AI agent, Copilot, a draft reply, a teammate by name, or Auto-summary for one Gleap condensed itself.
Mostly from AI, during conversations: something worth keeping gets written down. You can also add entries yourself on the Memory tab, edit any of them, or clear the lot.
Writing one by hand is the fastest way to fix a wrong assumption — rewrite the entry rather than hoping the next conversation corrects it.
Memory doesn't grow forever. Once a contact's memory reaches its limit, Gleap consolidates it — older entries are merged into summaries so the useful facts survive and the noise doesn't.
You can steer that with a custom prompt in the same settings, which says what consolidation should hold on to. If pricing tier and integration setup are what matter for your product, say so there.
The same goes for anything sensitive the customer shared once and shouldn't be repeated back at them later.
Disabling stops new entries being written and preserves what's already stored, so switching it off isn't destructive. To actually remove what's there, clear the memory on the contact.
