← All screens
P
Picaro
Home
Campaigns
Keywords
Dashboards
Automations
AI Agent
Agent 3
AI Agent
AACME JP
Sep 1 – Sep 9
3
Chat
Conversations
Insights
Knowledge
Billing
Proposed → approved
42 → 18
43% · healthy band
Median time to decide
4h 10m
71% within a day
Changes applied
16
3 undone
Measured effect
8 / 2
improved / worsened
All318 Decision chains42 Task runs96 Campaign changes64 Market9
Read-only record
Activity each decision from the insight that raised it to the effect it had
Sep 9
Lower bids on zero-conversion keywords
NEGATIVE_KW_REVIEW · SP-Auto-Discovery · APPROVAL_NEEDED
IMPROVED
INSIGHT 18% of spend on 12 terms with zero conversions 09:12
PROPOSED 3 keywords, −15% each · est. −¥2,000/mo 09:12
APPROVED Yamada · 2 of 3 items · 1 excluded as stale 09:14
APPLIED 2 bids changed · status partial_success 09:14
OUTCOME ACoS 42.1% → 33.8% · spend −¥1,940, sales flat +14d
CAMPAIGN SP-Autumn-Launch started by Sato · ¥4,000/day · agent holds off proposals for 14 days 08:00
TASK RUN DAILY_BUDGET_BURN · every_day · success · nothing to raise 08:00
Sep 8
Raise daily budget on SP-Brand-Protect
DAILY_BUDGET_BURN · APPROVAL_NEEDED
REJECTED
INSIGHT Out of budget by 14:00 for 5 days running 08:02
PROPOSED ¥8,000 → ¥11,000 daily 08:02
REJECTED Yamada · "close to the monthly cap" → agent stopped proposing this shape 16:20
MARKET Category CPC up 22% week over week · outcomes measured in this window report as can't tell 07:00
Sep 5
Add 4 negative keywords
NEGATIVE_KW_REVIEW · SP-Category-Broad
NO CHANGE
APPLIED Suzuki approved all 4 · success 11:30
OUTCOME ACoS moved −0.4pt · inside noise, treated as no change +14d
Did it work?
16 applied, measured 14 days after
Improved
8
No change
4
Worsened
2
Can't tell
2
Correlation with the change, not proof of cause.
How the agent has changed
learned from your decisions, newest first
Stopped proposing budget rises near the monthly cap
Sep 8 · from 1 rejection
Holds new campaigns out of proposals for 14 days
Sep 3 · account constraint
Never adds a brand term as a negative
Aug 28 · from brand guidelines
Bid steps capped at −25% per change
Aug 14 · from 3 edits before approval
What the agent reads from this
its own next steps, drawn from the record above
Re-test the SP-Category-Broad negatives
The Sep 5 change landed inside noise. A larger set would give a readable result.
Propose a portfolio cap instead of a budget rise
SP-Brand still runs out by 14:00, but budget rises get rejected near the cap.
Offer to automate the bid-down pattern
Approved 13 of 14 times, never edited, and the last three improved ACoS.