Autonomy
Agents that run the account.Within limits you wrote down.
Five specialists — structure, performance, budget, bidding, creative — read your Meta and Google accounts, propose changes, and apply them only inside the ceilings you set. Every change is typed, guardrail-checked, logged, read back and reversible. Start in dry run and promote them when a week of decisions has earned it.
Ripple is working.Why this exists
Ad accounts are optimized on a human schedule.
A few hours a week, a handful of decisions at a time. Between those hours, budget drifts, bids sit stale and a broken ad set keeps spending.
Ripple closes that gap without handing over the keys. The agents work the account continuously — and every move they want to make is checked against your rules before it reaches your ad platform.
Dry run
The agents read your account and decide what they would change. Nothing is sent anywhere.
- Every proposed change written out in full
- Checked against your guardrails before it is even queued
- You read a week of decisions before trusting one
Proposed change
Not appliedDaily budget · Search — Brand
$180$240
Logged with the reasoning and the rule it was checked against. Google never hears about it.
Shadow mode
The same agents, running the same decisions on a live cadence — recorded and scored, still not executed.
- Decisions time-stamped as they would have happened
- Scored against what your account actually did
- Guardrail hits shown, so you can tune the rules before they bite
Shadow result
SimulatedSame change, scored on 7 days
CPA $41.20$36.80
A record of what autonomy would have done, next to what actually happened.
Live autonomy
You turn it on, one guardrail at a time. Agents act inside the budget and the limits you defined — and only there.
- Typed, budget-bounded, guardrail-checked before it is sent
- Verified against the platform after the fact
- Reversible from the log, one change or the whole day
Applied change
Applied · verifiedDaily budget · Search — Brand
$180$240
Inside your daily cap and your maximum step change. Revert restores the previous value exactly.
A team of specialists, not one general model.
Each agent owns one part of the account and is judged on it. They share one view of your data, one set of objectives and one set of limits — so their proposals get reconciled before anything is queued, never fought out inside your budget.
The roster
Five specialists, one account
- Structure agentAudits campaign, ad set and keyword structure
- Performance agentReads results by objective, segment and window
- Budget agentMoves spend toward what is converting
- Bidding agentTunes targets against your cost ceilings
- Creative agentRotates and retires assets on evidence
Setup
Connect in minutes
Guided Meta and Google Ads OAuth — no copying access tokens between tabs. You pick exactly which accounts Ripple may see and which it may manage.
Objectives
You choose what wins
Leads, cost per acquisition, lead quality, return on ad spend. Rank them and every agent optimizes against that order instead of its own local metric.
Working hours
They do not wait for Monday morning
Analysis runs on a schedule, not on someone’s calendar. A broken ad set at 2am is caught on the next pass, not on the next weekly review — and the fix still has to clear your guardrails before it is applied.
Full autonomy. Never unrestricted.
Safety here is structural, not a promise in a settings page. A change that cannot be expressed, checked, bounded, logged, verified and undone simply is not a change the system can make.
- 01
Typed
A change is a value in a schema, never free text an agent improvised.
- 02
Guardrail-checked
Measured against your limits before it is even queued.
- 03
Budget-bounded
It cannot move more money than the ceiling you set allows.
- 04
Logged
What changed, why, and which rule cleared it — kept, in full.
- 05
Verified
Read back from Meta or Google after it lands, not assumed.
- 06
Reversible
Undo a single change, or roll the whole day back.
- Daily and monthly spend ceilings
- Maximum change per step and per day
- Campaigns and ad sets that are off limits
- Approval required above a threshold you choose
- Your objectives, ranked in your order
- Conditions that pause everything automatically
You write the guardrails once, in plain limits. Every agent, on every pass, is measured against them.
Change a limit and it applies from the next pass — no redeploy, no support ticket, no waiting for a human to reconfigure anything.
Connect an account.Choose who runs it.
Point your own assistant at Ripple and work in chat, or start the agents in dry run and promote them when a week of decisions has earned it. The guardrails are the same either way.
Meta and Google Ads OAuth · You choose the accounts · Disconnect at any time