Skip to content
ChatSonnet
Sonneto Optimize

Give every improvement a purpose.

Set a bot objective, record the starting point and compare the released candidate with its baseline. Connect repair decisions to observed outcomes.

Get started with Sonneto Read the setup guide
How it worksSee an exampleCurrent scope
BOT OBJECTIVESONNETO WORKSPACEExample

Every improvement starts with a goal.

01
ObjectiveTracked goal completion
02
BaselineFrozen at publication
03
Candidate reportCollecting relevant activity
Measure the release. Interpret with care.
Sonneto by ChatSonnet
01
UNDER THE SURFACE

Choose the outcome before the change.

A support bot and a lead qualification bot should not be assessed in the same way. Choose an objective that matches the journey, then check whether the corresponding events are recorded. A goal cannot be measured reliably when its tracking is absent.

Per-bot objectives

Select the available completion, tracked goal or handoff objective appropriate to your flow. Read metric definitions before deciding what success means.

Frozen release baseline

Publication captures a starting snapshot for the candidate release. Later comparisons should refer to that release rather than combine unrelated versions.

Collect enough activity

Reports with insufficient conversations remain in collecting status. A handful of successful conversations cannot establish a reliable improvement.

02
UNDER THE SURFACE

Treat a benefit as a hypothesis until measured.

Clearer wording may help visitors continue. Validate the candidate’s behavior first, then observe its release. A before/after difference alone does not prove that the repair caused it.

Reviewed improvements

Prepare wording and controlled connection changes with readable diffs and baseline-versus-candidate scenario results.

Narrow automatic changes

Exact owner-approved replacements and supported whitespace cleanup may qualify under a saved policy. General AI rewrites and flow changes still need review.

Learn from the result

Use the comparison to decide whether to investigate again. Proposed lessons remain editable and require approval before becoming trusted memory.

A PRACTICAL WORKFLOW

How it works.

  1. 01

    Set the objective

    Select a bot in Optimization and check its tracking.

  2. 02

    Release a tested candidate

    Prepare, review and publish a specific improvement.

  3. 03

    Compare the outcome

    Inspect the exact release report after sufficient relevant activity.

ILLUSTRATIVE EXAMPLE · NOT CUSTOMER RESULTS

From question to next action.

01Goal
Simplify a lead question without changing qualification requirements.
02Evidence
Illustrative evidence: one prompt combines several requests in a confusing sentence. Its commercial impact is still unknown.
03Action
Prepare clearer wording, compare the scenario results and publish after review.
04Follow-through
Compare tracked goals with the baseline. Review lead quality separately and account for traffic changes.
SCOPE & AVAILABILITY

What to expect.

Current Sonneto Optimize capabilities and conditions
ScopeDetails
AvailableObjectives, performance snapshots and post-release comparisons.
ConditionalRecurring work needs explicit bot enrollment and a functioning maintenance runner.
Separate roadmapBroad knowledge optimization and actual AI model fine-tuning.

Access depends on permissions, provider configuration and applicable AI allowance. Check current plans and pricing.

KEEP THE MOMENTUMExplore Sonneto Control & approvalsBuild your skills in ChatSonnet Academy
Sonneto by ChatSonnet
YOUR NEXT IMPROVEMENT STARTS HERE

Start with a question.
Build toward better outcomes.

One bot. One clear goal. A teammate to help you move forward.

Get started with Sonneto Read the setup guide
Compare plans and AI allowances
Independent.
Within your boundaries.Review first.
Expand when you’re ready.