Cut the losing bids
The day-one audit: every search term that spends and never pays, totalled, ranked, and drafted as a fix that waits for your approval.
One endpoint gives your agent 55 tools over your live campaigns and your real subscription revenue.
agentWant me to schedule this sweep every morning?
Every one of these is a question your agent can answer through the same endpoint, against your live account and your real subscription revenue.
The day-one audit: every search term that spends and never pays, totalled, ranked, and drafted as a fix that waits for your approval.
Suggestions ranked by what your paying users actually search, not raw volume. One word adds the winners as exact-match.
Apple counts installs. SAR joins your real subscription revenue, so ROAS means dollars, and the budget follows the truth.
Not charts: actual customers. Ask any keyword who it brought in, what plan they bought, and every dollar they’ve paid since. Tracked, not modelled.
Say it once and it becomes a Monday ritual, scheduled inside your own agent: the report lands in your chat and the rebalance is drafted. With a read token the approval stays yours.
A change isn’t “done” until Apple confirms it. Watch one move through every stage (proposed, your approval, applied, verified) and see the live status your agent can read at any moment. On the plan flow nothing touches Apple until you say yes. A write scoped token is the deliberate exception: it lets your agent act directly.
Nothing has touched Apple yet. Two changes wait for your yes.
write-safety: validate → apply → read-after-write.
Re-read after applying, Apple confirms:
Your agent can read and reason about your ads all day, but it can’t reach into your account to change anything. That lives behind a login it was never given. SearchAds Radar is the bridge: it holds the connection, exposes 55 tools over MCP, and turns intent into real, verified changes: approved by you on the plan flow, applied directly if you issue a write token, and logged either way.