Frequently asked questions

An agent can’t make a change past your hard limits. Spend caps, the 20% limit on each budget change, contact caps, consent and suppression lists, and banned phrases are checked by software when an agent proposes a change, and again right before it’s made. Claims are checked by the brand review agent against your approved claims library, and a claim without a source can’t be used. Every action is logged with its reason, and one switch stops every agent.

The ones you choose. Every kind of action has its own permission level: You approve, Acts and notifies you, Acts silently, or Not allowed. Budget and bid changes, pausing and resuming start at Acts and notifies you. Launches, sends, using library creative in ads and messages, and audience changes start at You approve. Change any of them, for every agent or just one.

Each job uses the model that tested best for it, chosen from several providers. Your rules, history and library don’t depend on the model, so a better one can be swapped in without changing how you work.

Pricing isn’t set yet. Your ad spend goes straight to the ad platforms, as it does now.

The platforms’ automated campaigns are good at finding buyers inside their own platform, from its data and your product feed. But each one sees only its own platform and grades its own results. Agents here work across the channels you connect, email included, under one set of rules you control. They judge results on your own conversion data and give a reason for each decision. To an agent, an automated campaign is a campaign like any other: it can propose budget, pause and resume changes for it.

Your email platform’s agent works inside that platform, on its numbers. Agents here run your ads and your email together, under one set of limits, and are judged on your own conversion events.

We don’t take the ad platforms’ word for it. Each platform credits itself for any sale it touched, so their totals overlap and run high. Agents measure against your own sales data instead, compared with a baseline from your history. By default, each email send holds back a small group, so across sends you can see what email adds. For ads, when you have the volume, agents propose tests that leave out a region or an audience. When there isn’t enough data to tell, the report says so.

Agents are judged on your own conversion events, from your store, your CRM or the events API. Each event carries what it’s worth to you, such as order value or margin. Each platform’s count still guides its own bidding; decisions across channels use your events.

No. You connect each account through the platform’s own sign-in (OAuth), so your passwords never reach us. Agents never hold credentials of any kind: they propose changes, and a separate service makes them after checking your rules. You can revoke access at any time.

Each company has its own workspace. Its agents act only through that company’s rules and the service that makes changes, and they never hold the credentials to your accounts. Every action they take is logged in that company’s audit record.

Standard API access to your ad and email accounts, granted through each platform’s own sign-in. Agents never hold those credentials. They propose changes, and a separate service makes each one after checking your rules again.

A wide range across ads, email, store and CRM, each connected through the platform’s own sign-in, plus an events API for anything else you count, such as a booked demo or a signed contract.

Two of the agents: the creative desk and the copy desk. They draft images, headlines and emails from your brand and approved material, and brand review checks each one before it goes out. No draft runs until it’s approved. That’s you, unless you let a desk approve work for other agents.

At the start, connect your accounts, agree each agent’s first brief and choose what needs your approval. After that, it’s mostly reading: the morning reports, the few decisions in your inbox and the occasional question from an agent. Loosen permissions as an agent earns your trust.

It depends on how fast your sales come in. A store with orders every day can call a test in two weeks. A business that sells through leads may wait weeks for qualified leads and closed deals, so its agents act less often and judge on earlier steps. Until there’s enough evidence, the report says it’s too early to call.

No, a CRM is optional. Conversion events can come from your store, your CRM or the events API. If you sell through a sales team, sending stages such as qualified lead and signed contract lets agents judge on more than form fills. Without them, agents judge on the steps your site records.

Photos, yes. Upload your own, and agents work only from approved ones. The creative desk makes still images, and video you upload is kept beside them.

Agents stop acting at once. Your campaigns, audiences and emails live in your own accounts and keep running at their last settings. Your record of every action, with who proposed it, why and who approved it, remains yours.