Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.meshpilot.app/llms.txt

Use this file to discover all available pages before exploring further.

The Ads agent is the deepest-built agent in Mesh Pilot today. It runs against your connected Meta, Google, TikTok, and Amazon ad accounts.

​
What it does

The Ads agent polls your ad-platform APIs on a regular cadence (typically every few hours) and pulls down spend, revenue, ROAS, frequency, CTR, and audience signals for every active campaign, ad set, and ad. It then runs that data through a small set of rules — anomaly detection, creative-fatigue scoring, budget-vs-target reconciliation, audience-drift checks — and writes one observation per useful event into the Brain.

​
What it proposes

Common proposals you’ll see in the Inbox:
  • Pause an underperforming ad set that’s burned past its CPA target.
  • Scale budget on a campaign that’s outperforming its ROAS target with headroom.
  • Refresh creative when an ad set’s frequency crosses a fatigue threshold (often paired with a Creative-agent loop).
  • Adjust bid strategy on a campaign whose delivery is misaligned with its objective.
  • Investigate an unusual spend or ROAS swing — surfaces a one-shot proposal with the root-cause analysis attached.

​
What you approve

Each proposal shows the agent’s reasoning, the data window it acted on, the predicted impact (e.g. “saves ~₹12k/day at current spend”), and the exact change it will write back to the platform. You can approve as-is, edit the payload (e.g. tweak the budget number), or reject with a one-click feedback signal.

​
What the Brain learns

Every approval, edit, and rejection feeds the agent’s future proposals. Patterns it accumulates and shares — anonymised — across brands include creative-fatigue thresholds by vertical, time-of-day delivery patterns, audience-overlap warnings, and budget-pacing rules of thumb.

Open the Ads surface

See live ad state and recent agent activity.