Skip to content
← Insights

Ask any action why it exists.

Every proposed and executed action carries an explanation written for a person rather than a log reader: what changed, why now, what it is expected to do and what happens if it is wrong. Attached to it is the evidence bundle — the metric and attribution queries that ran, the truth layer each figure came from, the Constitution clauses checked and the approval recorded. The explanation is written over that bundle, so it cannot describe evidence the action never had. Ask a follow-up and the answer is drawn from the same record.

Generated from the record

The model writes over the stored inputs and tool results, not from a general impression of the account. Where a number appears in the text, it is the number the deterministic service returned at decision time. An action with thin evidence gets a thin explanation, which is exactly the signal you want.

Reversal is part of it

Each explanation states the previous state and the path back to it. Budget moves, audience changes and status changes are reversible, and the text says how to reverse them. Anything that cannot be undone is flagged before approval, not after.

Answers a question, not a diff

'Why did the Meta budget rise on Tuesday' returns the evidence read, the Constitution clause that permitted the move and who approved it. Finance, an auditor and whoever inherits the account read the same text. Nothing rests on someone remembering what the team decided that week.

In this capability

  • Per-action evidence bundle
  • Plain-language rationale
  • Stated rollback path
  • Follow-up questions answered

One goal. Every channel. Real business truth.

Connect your ad accounts, say what a good customer is worth, and let the Growth Brain work the objective inside limits you set.

No ad spend commitment · Connect one channel to start · [email protected]