How AdAmigo.ai Autopilot works

Autopilot is the part of AdAmigo.ai that works on your ad account every day without being asked. Every morning at 3 am (your ad account's time zone) it reviews your Meta and Google accounts and prepares the changes it would make, each with what it wants to do, why, and the expected impact. You approve each change with one click, or turn on auto-apply and let it run your account within the goals and limits you set.

Diagram of what AdAmigo.ai Autopilot does every day: a 3 am daily audit leads to recommendations, which you apply with a click or with auto-apply, and the change is made and fed back into the next day's audit.

Think of it this way: you hired an AI marketer to run your paid ads. In the chat agent, you lead: you say what you want and it does the work. Autopilot is the other half. It comes to you. Every day it brings you the changes it would make, explains each one, and waits for your OK. Nothing happens in your account until you say so, unless you choose to turn on auto-apply.

What does Autopilot do every day?

Every day follows the same cycle:

  1. 3 am: the daily audit. Autopilot reviews your account: what happened yesterday, your history, the season, your competitors and, if connected, your stock levels.

  2. Recommendations. It turns what it found into that day's recommendations. A recommendation is one change it would make, like raising a budget or pausing an ad that isn't working. Each one says what it wants to do, why, and the expected impact.

  3. You decide. By default you click Apply on each one you agree with. If you've turned on auto-apply, they're applied for you, always inside your limits.

  4. The change is made. It happens in your ad account close to real time. You can see it right away in Campaigns in AdAmigo.ai, or in Meta Ads Manager and Google Ads.

  5. Results feed back. Autopilot measures what each change did and uses it the next day. So tomorrow's recommendations are better than today's.

That last step matters most. Every change teaches it more about how your account behaves. Read more in how AdAmigo.ai learns your account.

Where do you find Autopilot?

Autopilot sits in the side panel on the right, next to the chat. Three clicks show you everything.

Step 1: Click Autopilot in the side panel

The red number on Autopilot shows how many recommendations are waiting for you.

The AdAmigo.ai app with the Autopilot tab in the right side panel highlighted and magnified, showing a red badge with the number of open recommendations.

Today's recommendations open under Needs your approval.

Step 2: Click More info on a recommendation

Each card has a title and a one-line summary. Click More info to see the full reasoning.

An expanded AdAmigo.ai Autopilot recommendation to launch 2 new ad creatives for a Summer Collection, with the What, Why and Expected impact sections marked 1, 2 and 3.

The card opens into What, Why and Expected impact.

Step 3: Click 4 Active

The button top right shows how many agents are active. Click it to see them.

The AdAmigo.ai Autopilot agents list opened from the 4 Active button, with the button marked 1, the Strategist's lightning bolt for auto-apply marked 2 and its pencil icon marked 3.

The agents list opens, with a lightning bolt (auto-apply) and a pencil icon next to each agent.

Autopilot also has four tabs at the top:

  • Today: this day's recommendations, split into Needs your approval and Scheduled to apply.

  • Scheduled: changes and custom automations set to run at a set time, for example every day at 7:00 AM.

  • Implemented: changes already made in your ad account.

  • Blocked: changes that couldn't be made, for example an ad Meta disapproved for a policy reason, with an explanation of what happened.

For every button on a card, see managing Autopilot actions.

Who are the agents?

Autopilot is a team of agents, each with its own job. Every recommendation shows which agent made it (top left of the card) and the platform it's for, Meta or Google (top right).

  • Strategist: sets the overall plan for your account, based on your goals and priorities.

  • Media Buyer (Meta): makes the day-to-day changes in your Meta ads (Facebook and Instagram).

  • Media Buyer (Google): makes the day-to-day changes in your Google ads.

  • Creative Designer: makes new versions of your ads, mostly new takes on ads that already work.

  • Protect: keeps your account safe, for example by flagging an ad that was disapproved. More on the Protect page.

  • Creative Strategist: coming soon.

Each agent shows as Active or Paused, and you can pause or restart it. Auto-apply is set per agent, with its own lightning bolt. Below the agents you'll find Custom Automations, for jobs you define yourself, like a daily report. Those are covered in custom agents.

What does Autopilot know about your account?

This is what makes Autopilot different. Before it recommends anything, it does the homework that no one on your team has time for. Every recommendation is based on the research in your knowledge base, and more:

  • Your ad account history and the experiments you've run before.

  • Seasonality and trends, down to which days of the week are strong or weak for you.

  • How your account reacts to changes, for example what happens to results when a budget jumps.

  • Market research: your ideal customers, your competitors, their ads and messaging angles, and where you're stronger or weaker than them.

  • Your priorities and guardrails, the goals and limits you set.

  • Your connected tools, for example stock levels from Shopify.

The expected impact on each recommendation isn't a guess. It's modeled from historical analysis of how certain changes affect performance. Autopilot uses it to decide what to do first and to manage risk: changes with a high chance of hurting performance are left out.

Here's the plain truth. Hardly any AI tool, and very few people, do this kind of homework on how changes affect results. Most tools follow fixed rules. Most people go on a hunch and whatever they remember from last month. AdAmigo.ai collects these data points at scale on your account and makes every decision based on them. Three examples show what that looks like.

Weak weekends

Say weekends perform badly in your account. Autopilot won't start a creative test (a fair comparison of new ads against each other) on a Friday. The test would run straight into your weakest days and make the new ads look worse than they are. It starts the test on a better day instead. More in creative testing tips for AI-managed campaigns.

Budget jumps

Raising a budget by a lot more than 20% at once doesn't just restart Meta's learning phase (the period when Meta is still working out who to show your ads to, and results bounce around). Our research shows it can hurt performance over the following weeks. So Autopilot scales your budgets in a way that's proven for your account. More in how to scale Meta ad budgets without losing ROAS (ROAS is return on ad spend: how much you earn for every dollar you spend on ads).

Stock running out

With Shopify connected, Autopilot sees when a product is about to sell out. It can then recommend lowering spend on ads for that product, so you don't pay for clicks on something people can't buy. See connectors to link your store.

What can Autopilot change?

Pretty much everything you'd otherwise do by hand in Ads Manager:

  • Build new campaigns.

  • Build new ad sets (on Google: ad groups). An ad set is the group inside a campaign that holds the budget, audience and ads.

  • Design new ads, mostly new versions of ads that work. For example, to keep a new account supplied with fresh ads, or to replace a winning ad that has started to wear out because people have seen it too often (creative fatigue).

  • Change budgets.

  • Change bidding, meaning how much you're willing to pay for a result.

  • Change targeting, meaning who sees your ads.

  • Add keywords and negative keywords on Google. Negative keywords are searches you don't want your ads to show up for.

  • Run creative tests. It spots new ads added to the account, sets up the right way to test them, pauses the losers and scales the winners.

The upshot: you don't need automated rules, custom rules or hands-on work in Ads Manager anymore. Rules only do what you told them last month. Autopilot looks at your whole account every day and explains every move. See AI vs rule-based automation for the full comparison, or the AI Actions product page.

Why doesn't it change everything on day one?

Autopilot works like a chess player. Sometimes the big move needs two or three smaller moves first.

For example, in the first 2 or 3 days it may only lower budgets and pause what isn't working. That makes the account more profitable and frees up money. Then on day 4 or 5 it may use that money for its first new campaign. If you only looked at day 1, you'd see a few cuts. Looked at over the week, it's one plan, played out in order.

Why are there no recommendations some days?

Because nothing needed doing. Autopilot only recommends a change when something needs to change. It doesn't make changes just to look busy.

That's on purpose. Too many changes hurt ad accounts: tests need time to play out, and editing too often resets the platforms' learning, so results get worse instead of better. Autopilot keeps that in mind. But when something does need doing, it gets done that day, not three days later when someone remembers.

On your very first day, recommendations can take a few hours. Before it suggests anything, Autopilot runs a deep analysis of your account and market and builds a strategy. At the latest, your first recommendations are there the next morning.

Does Autopilot change things without asking?

No. By default nothing changes until you click Apply. Read the recommendation, click Apply, and the change is made in your ad account. The card then shows Applied just now.

An AdAmigo.ai Autopilot recommendation card showing Applied just now, highlighted and magnified, after the Apply button was clicked.

It happens close to real time, so in the beginning you can cross-check everything it does. Click Campaigns in the side panel to see the change, or open Meta Ads Manager or Google Ads.

The AdAmigo.ai Campaigns tab in the side panel with a Skincare Conversions campaign row highlighted and magnified.

You're not limited to yes or no, either. On any card you can Ask a question about it, Edit it before applying, or delete it with the trash can icon and say why. That feedback teaches Autopilot the things that live in your head and not in the account. Most people delete a few recommendations in the first week, and soon find themselves approving almost all of them. See managing Autopilot actions and using Campaigns.

That's the moment to turn on auto-apply. Click the lightning bolt next to an agent, and its recommendations move to Scheduled to apply, showing Applies tonight at 12:00 AM. You can still read, ask, edit or delete them before then, and every change stays inside your guardrails. Auto-apply is always your choice, never the default.

What's left for you to do?

Very little, and only the parts that need you:

  • Strategic pivots. A new product line to push, or a new country to open. Add it as a priority: click your name in the bottom left, then Settings, then Priorities.

  • Budget limits, when they change. Update them in Settings under Guardrails, for example the Maximum daily spend. If a priority and a guardrail ever clash, the guardrail wins.

  • Launching ads made somewhere else, for example by a designer or an influencer. Send them through the chat agent, as shown in bulk uploading and launching ads.

Everything else runs on Autopilot. More on setting goals and limits in priorities, guardrails and memory.

Frequently asked questions


What is AdAmigo.ai Autopilot?

Autopilot is the part of AdAmigo.ai that reviews your Meta and Google ad accounts every day and prepares the changes it would make. Each recommendation explains what it wants to do, why, and the expected impact. You approve each one with Apply, or turn on auto-apply so changes are made for you within your limits.

When does Autopilot create new recommendations?

Every day at 3 am in your ad account's time zone, Autopilot runs its daily audit and creates that day's recommendations. You'll find them under Today in the Autopilot tab of the side panel. With auto-apply on, recommendations are still created at 3 am and then applied at the scheduled time shown on each card.

Why don't I see any recommendations yet?

On your first day, Autopilot runs a deep analysis of your account and market and builds a strategy before it suggests anything. That can take a few hours. At the latest, your first recommendations are there the next morning. On later days, an empty list simply means nothing needed changing that day.

Will Autopilot change my ads without asking?

No. By default nothing changes until you click Apply on a recommendation. You can also ask about it, edit it or delete it first. Auto-apply, where changes are made for you, is off until you turn it on per agent with the lightning bolt, and it always stays inside your guardrails.

What can Autopilot change in my ad account?

Autopilot can build new campaigns and ad sets (ad groups on Google), design new versions of ads that work, change budgets, bidding and targeting, add keywords and negative keywords on Google, and run creative tests. In tests it pauses the losing ads and scales the winners. Every change is shown first with what, why and expected impact.

Do I still need automated rules?

No. Automated rules only do what you told them in advance, like "pause if cost per result goes above a set amount." Autopilot looks at your whole account every day, including history, seasonality, competitors and stock, and explains every change it proposes. You don't need custom rules or hands-on work in Ads Manager anymore.

Why does Autopilot sometimes recommend nothing?

Autopilot only recommends a change when something needs doing, and never changes things just to look busy. Too many edits hurt ad accounts: tests need time to play out and frequent changes reset Meta's and Google's learning. When something does need doing, it recommends it that same day.

How does Autopilot decide what to do first?

Each recommendation has an expected impact, modeled from historical analysis of how certain changes affect performance. Autopilot uses that to rank what matters most and to manage risk, leaving out changes with a high chance of hurting results. Your priorities and guardrails, set in Settings, shape every decision too.

Next steps

  1. Managing Autopilot actions

  2. Auto-apply

  3. Priorities, Guardrails and Memory

New here? Start your 7-day free trial. Already signed up? Open the app and click Autopilot in the side panel to see today's recommendations.