ATRIBU AGENT

Every account audited. Every fix priced. Every change verified.

The agent works your Meta and Google accounts overnight and hands you decisions, not dashboards. Nothing changes without your approval — and everything that changes is checked and logged.

Plans from $99/mo · Agency (10 client workspaces) $399/mo — every plan starts with a walkthrough

TODAY’S RECOMMENDATIONS · 3

Pause “TOFU Follow Me — Broad”

LOW RISK · REVERSIBLE

CPA has sat at $210 against a $60 target for 3 straight days — 41 clicks this week, 0 qualified leads. The rest of the campaign is unaffected.

// preview — what confirm will do

1 Meta call · POST /{ad_id}

status: ACTIVE → PAUSED

saves $38/day · resume anytime

Approve & executeDismissre-verified on Meta at T+5 min

THE PROBLEM

Ads Manager shows charts. You still do all the thinking.

01

Six tabs deep, every morning

Scanning accounts for drift is an hour a day per client — and the important change is always the one you didn’t catch.

02

Gut-feel changes, no receipts

Budget moves happen on instinct, with no price tag, no confidence score, and no record of what was changed or why.

03

Click-trained decisions

Platform metrics count clicks. Your bank counts payments. Decisions made on the first rarely maximize the second.

THE LOOP · EVERY 24 HOURS

Propose → approve → execute → verify.

1 · It proposes

Overnight audit finds drift and opportunity; each finding arrives priced in $/day with a confidence score and a risk tier.

2 · You approve

Preview shows the exact API call before it happens. Only workspace admins can confirm. No approval, no change — ever.

3 · It executes

The change goes live on Meta within seconds, with the account’s pre-change state captured first.

4 · It verifies

Five minutes later it re-checks Meta to confirm the change landed as intended — then writes the immutable audit row.

SAME-DAY REPLAY PROTECTION — approving the same action twice never double-writes

WHAT IT CAN TOUCH

Four moves. Clamped, gated, logged.

The agent’s write access is deliberately narrow. Everything else — targeting, creatives, campaign structure — stays read-only.

Pause an underperformer

When an ad drifts past its CPA target with enough data to be sure, it proposes the pause — priced as $/day saved.

1 Meta call · status=PAUSED · low risk

Scale a winner

High attributed ROAS plus healthy frequency earns a budget increase — never more than the guardrail allows.

daily_budget +10–50% clamp · medium risk

Reallocate between ad sets

Moves budget from a laggard to a leader in one approved action — the down-shift and up-shift execute together.

2 Meta calls · 10–35% clamp · medium risk

Refresh a fatiguing creative

Fatigue predicted before CTR collapses. One click hands you the ad’s brief and performance context — ready for your designer, or your AI.

no Meta call · brief handed to you

Why its advice is worth taking: the agent reads payments, not platform metrics. Every sale in Stripe, MercadoPago, or your CRM is traced back to the ad that caused it — so “winner” means money in, not clicks out.

How the agent works — read the docs

RUNNING AN AGENCY?

One briefing across every client account.

The Agency plan watches up to 10 client workspaces — findings arrive in a single morning brief, tagged per client, with client-ready reports on schedule.

Book a demo
Tomás Parraguez

“We needed real ROAS and real CAC — not platform numbers. Atribu gives us a decisive view of what to scale and what to kill.”

Tomás Parraguez · Founder, Markgrowth & Markdroid · @tom_parraguez

The first briefing lands tomorrow at 07:58.

Connect Meta on the call — the agent starts its first shift tonight. Demos in English or Spanish.