New AI Agent can now build your knowledge base, connect channels and invite agents — all by chat Try it now
Help homeProductPricingBlog
Start freeLog in

Manage skills and customer memory

Inspect learned capabilities, disable incorrect skills, and manage customer-specific memory.

Updated 2026-08-30

Walk through the demoDemo data
OverviewBuilt-in AI is handling reception

Check whether AI is safe to serve, what it answers from, and what needs you.

AI capabilities & learningManage knowledge & skills →
12Data sourcesReindexed in the last 2 hours
64Skills5 skill matches in the last 7 days
174Learning suggestions22 adopted · 7 used
2Customer memory1 low-confidence memory to watch
Last 7 daysView reports →
35Conversations
35Closed
0Queued
35Assigned
Daily AI-handled conversationsConversations where AI actually repliedUnit: conversations
6/146/156/166/176/186/196/20
Proactive outreach weekly reportReview weekly trigger decisions, cooldown blocks, and delivery failures.2 cooling down
5Sent8Eligible4Blocked1Delivery failed

3 outreach tasks open

Needs your attentionReview suggestions →
How should we reply when sizing runs small?A staff answer has been turned into a skill draft and needs confirmation.Handle
What if DHL takes more than 72 hours?A staff answer has been turned into a skill draft and needs confirmation.Handle
Learning result delivery failedA Telegram learning result matched once but did not reach the customer. Review the failed replay or retry from Learning review.View
What AI recently learnedView all learning →
UsedHow do I track a shipment?

Look it up using the order number.

Last 3 real matches: Web Widget · Custom API — 1 externally delivered, 2 pending verification, 0 failed.
5 matchesView learning recordTest/manage skill
Step 1 / 5
Open AI skills

Open AI skills under the AI sidebar's learning and assets section to inspect adopted and generated capabilities.

Skills and customer memory solve different problems. Skills are reusable business capabilities. Customer memory adds context for one specific customer.

Skills#

Open Workspace, select Skills, and inspect adopted and generated capabilities. A skill detail can show:

  • trigger or applicable customer question;
  • response steps, required business information, and handoff boundaries;
  • learning source and version;
  • tests, runs, and real usage;
  • current enablement state.

Skill details explain how the skill handles a request without requiring merchants to configure internal processes. If a skill is wrong, pause or roll it back, then create a corrected version through correction and an improvement suggestion.

Customer memory#

Customer memory stores preferences, restrictions, or durable context for one customer, such as language preference or a material allergy. It does not become a skill shared with every customer.

Open Customer memory from an adopted improvement suggestion, a skill source, or the customer profile. Repeated statements of the same preference are merged for that customer. A clear and newer preference automatically replaces the old value while preserving version history. If a new statement is ambiguous, the system keeps the current trusted value instead of letting uncertain content overwrite it or waiting for team approval. Incorrect memory can still be edited or removed.

Customer memory cannot override policy facts in the knowledge base and should not contain unnecessary sensitive data.

How to choose#

Use skills or knowledge for behavior shared across customers, customer memory for one customer's preferences, and Yuna long-term memory for the merchant team's own working preferences.

Didn't find your answer?

Start free and try the real product, or reach out — humans are as online as the AI.