SourceOwner-reviewed feedback
Sonneto Learn
Build understanding you can correct.
Give Sonneto approved context about your workspace and bots. Review proposed lessons, keep sources visible and update what it remembers.
A memory your team can inspect.
ScopeWorkspace or selected bot
StatusProposed · awaiting review
01
UNDER THE SURFACE
Make business context explicit.
Chat history records an interaction. Learning memory stores context deliberately approved for reuse. Owners can add information, scope it to a workspace or bot, review proposed text and correct it later.
Scope matters
Use workspace memory for shared preferences and bot-specific memory for one journey. A preference for one bot should not silently govern the rest.
Keep a source
Record where the information came from and check that it remains current. Edit proposed wording before approval and remove obsolete facts.
Review proposed lessons
Configured diagnosis can propose lessons from supported evidence and outcomes. The interpretation remains a proposal until an owner approves it.
02
UNDER THE SURFACE
Teach context without turning guesses into policy.
A visitor message is evidence of what someone said or needed. It is not an authoritative business rule. Verify the source before approving a price, policy, service condition or other material fact.
Approval before trust
Proposed lessons do not automatically become trusted memory. Reject incomplete inferences or edit them into a supported statement.
Maintain the source of truth
Keep changing product and policy information in the authoritative knowledge source. Memory should complement that source rather than become an unchecked copy.
Understand the mechanism
Approved memory supplies context to supported analysis and agent work. It does not retrain the provider’s AI model or change its weights.
How it works.
- 01
Inspect memory
Open Learning and review existing information or proposed lessons.
- 02
Verify wording and scope
Check business accuracy, source and the intended workspace or bot.
- 03
Approve and maintain
Approve useful context and edit or remove it when circumstances change.
From question to next action.
- 01Goal
- Remember the owner’s preferred order of qualification questions.
- 02Evidence
- Illustrative owner feedback: ask about the project before requesting contact details.
- 03Action
- Create a bot-scoped preference with the owner’s feedback as its source; edit and approve it.
- 04Follow-through
- Later supported analyses can use that preference. The live bot still needs its separate repair and publishing process.
What to expect.
| Scope | Details |
|---|---|
| Available | Editable, scoped, owner-approved memory and learning proposals. |
| Review required | Proposed lessons need approval before becoming trusted context. |
| Not included | Training on every conversation or fine-tuning model weights. |
Access depends on permissions, provider configuration and applicable AI allowance. Check current plans and pricing.
YOUR NEXT IMPROVEMENT STARTS HEREStart with a question.
Start with a question.
Build toward better outcomes.
One bot. One clear goal. A teammate to help you move forward.
Compare plans and AI allowancesIndependent.
Within your boundaries.Review first.
Expand when you’re ready.
Within your boundaries.Review first.
Expand when you’re ready.