Autonomy Level 2, Autopilot: the process starts on its own and reports back at the end
"Business on autopilot" is a promise you'll find in plenty of courses, blog posts and books. What's usually meant is a business that runs without the owner being needed at every turn. We share that goal. We describe the road there in smaller and more precise terms: with us, one process after another goes on autopilot. And Autopilot is a clearly defined level: Autonomy Level 2. This article covers what it means, what it doesn't mean and how you can tell that a process has reached it.
Why not the whole business at once
A business with a team has a lot of recurring processes. Each one has its own variants, its own way in and its own spots where something can go wrong. An AI employee that takes over a process like that gets trained: first, we watch your team at work with you, then every variant is shown once, with someone sitting next to it. That takes people from your team, and they have their actual work to do on the side.
That's why we tackle the processes one after another. It all starts with a map of every recurring process in your business. From that map, we pick the process that saves the most time or money, or brings in the most revenue. Only once it's running does the next one follow. "Business on autopilot" is then the sum of many processes, each of which was put on autopilot on its own.
What Autopilot means for us
A process is at Autonomy Level 2, Autopilot, when three things are true: it starts on its own on a schedule or trigger, it runs through to the final report, and it hands anything uncertain to a person. The work is done by an AI employee that has a role, fixed rules and clear limits. What an AI employee actually is, and how it differs from a chatbot, Kevin explains in What Is an AI Employee?
It starts on its own
The trigger is either a fixed schedule, for example every morning or every 30 minutes, or an event, for example an incoming email. Nobody on your team has to give it a nudge. That's exactly what sets Level 2 apart from Level 1, where someone starts the AI employee and the team checks every run. Which trigger applies is written down in the process profile, along with start, work, end and report. How a profile like that comes together is covered in Documenting processes so an AI employee can take them over.
It runs through to the end
A process on autopilot runs without a nudge up to the point where you can tell it's finished and something has been produced: a filed document, a prepared reply, an updated list. Then it sends its report.
It hands anything uncertain to a person
If the AI employee is unsure, it sends the case to a person on your team. In a feedback session, the AI employee is trained on the new variant, and it's saved as a rule. We call this the escalation path. It makes sure the AI employee doesn't guess on a case it doesn't know yet. And it makes sure the same case goes through without a person next time.
What Autopilot doesn't mean
The word conjures up pictures that aren't accurate. We'd rather clear four of them out of the way right now:
- Unsupervised. A report comes after every run. And if the report doesn't show up, that's the alert: something isn't running, and someone takes a look.
- Without an escalation path. A process without an escalation path doesn't get to Level 2 with us. The escalation path is part of the process profile, just like trigger and end.
- Without your team. Your team answers the cases from the escalation path, trains the AI employee on them and signs off wherever a person has to say yes. The goal is that your team only approves.
- At the push of a button. A process only goes on autopilot after training and after several clean runs. Until then, it's at Level 1.
What the final report contains
The report is the part that decides whether a process really takes work off your plate. Without a report, you end up following up yourself, and you're back to managing every step. A report you have to read in full to know whether anything needs doing just moves the work around.
This is how a completion email is structured in Owner Autonomy:
- The subject line carries the numbers: process, date, done, open, exceptions. You can tell right in your inbox whether you need to open it.
- Done: how many items, ideally broken down by type.
- Open: what didn't get finished and why, each with a note on where the item is or who it went to.
- Exceptions: anything that left the normal path, for example a case that went to your team via the escalation path.
- Needs a decision: what only a person can decide. If there's nothing here, the AI employee writes "none". Then you know it thought about it.
The bar is simple: after 30 seconds with the email, you know whether you need to do something. If not, the structure gets sharpened.
Some things can't wait until the end. For those cases, it's defined when the AI employee reports immediately: when a run can't start or breaks off, when access is missing, when the volume is unusual, or when a step is coming up that can't be undone. The AI employee only carries out a step like that when a person explicitly signs it off, every single time.
Who gets the report is up to you: you, the person on your team who looks after the AI employees, or both of you. The channel can be simple. An email is plenty to start with.
How a process gets onto autopilot
Every process starts at Level 1. That's where the AI employee is trained, in four steps: watch how the team does it. Sit alongside and show every variant once. Remember and ask, meaning use the escalation path and train in the feedback session. And finally, sign off. What happens at Level 1, and when we keep a human involved for good, is covered in Autonomy Level 1, Assisted: when your team checks every run.
The sign-off for Level 2 has a fixed threshold: four clean runs. After that, your team watches alongside for a while, and only then does the process run on its own with a final report. Clean means: the run went all the way through, the result is right, and nobody had to step in.
Autopilot is the level our guarantee covers
We give a guarantee for processes no. 1 and no. 2, and it measures exactly this level. Word for word: if your processes no. 1 and no. 2 aren't running on Autopilot after 90 days, Kevin keeps working with you at no extra cost, as long as you keep the agreement:
- You come to the 1:1 every week.
- You complete the agreed actions before the next 1:1.
- You name one person on your team who looks after the AI employees with you.
- You post your progress in the community every week.
Autopilot here means: the process starts on its own, on a fixed schedule or triggered by an event, and runs through to the final report without a nudge. That's Autonomy Level 2. Within the first two weeks, we agree in writing which processes are no. 1 and no. 2. Anything beyond these two processes and this level is a goal, not a commitment.
What this looks like for us
At a tax advisory firm, two processes are on autopilot. The first is filing digital tax assessment notices, up to 30 to 40 a day. An AI employee for filing handles that. The second process covers notices that arrive on paper. For it, a second AI employee checks the inbox every 30 minutes, spots the emails with paper notices and hands them over to the AI employee for filing.
Put the three features of Level 2 next to it, and each one is met:
- It starts on its own. Nobody in the office team kicks off filing. For the paper notices, the trigger is a fixed schedule, the check of the inbox every 30 minutes.
- It runs through to the end. At the end, the notices are filed, and filing reports back with a completion email.
- It hands anything uncertain to a person. Anything the filing employee can't match with certainty goes to the office team by email, and the AI employee is trained on it afterwards.
Both processes got there along the road this article describes. First, Kevin watched the office team: which types of assessment there are and how each one gets filed. Then the AI employee was trained, one type of assessment at a time, with someone sitting next to it. Filing was signed off after four clean runs. After that, the team watched alongside it for a while. And because incoming work passes through two steps, the two AI employees hand the notices off to each other, from the inbox to filing. Both processes are at Autonomy Level 2, Autopilot (as of September 2026).
The office team hasn't gone anywhere. It answers the cases from the escalation path, and each of those cases gets trained in afterwards. That's exactly what a process on autopilot looks like: it runs on its own, and the team only approves where it's actually needed.
What comes after Autopilot
Once the first process is on autopilot, the next one on the map is up. But every additional process at Level 2 also brings another completion email. At some point, the pile itself becomes work again. Then the next step is Level 3: an AI manager reviews the work, bundles the reports into a morning report, and you only see the exceptions. How that works and who checks then is covered in Autonomy Level 3, Audited: who checks the AI when you no longer do? The overview of all four levels is in The four autonomy levels: how a process in your business runs on its own.
Frequently asked questions
On autopilot, does my whole business run without me?
What's on autopilot is always one single process. A business where the most important processes run on autopilot and your team only approves is the goal. You get there one process at a time.
How can I tell a process is ready for autopilot?
By four clean runs in a row and a period in which your team watched alongside without having to step in. Until then, it stays at Level 1.
What happens if a run fails?
For that case, it's defined that the AI employee reports immediately when a run can't start or breaks off. If it fails in a way that it can't report anymore, the completion email doesn't show up. That's an alert too, and someone on your team takes a look.
Who gets the final report?
You decide that for each process: you, the person on your team who looks after the AI employees, or both of you. Later, at Level 3, you get one morning report instead of many emails.
Which process would be no. 1 for you?
If you're thinking of a process that runs the same way every day and still ties up someone on your team every time, that's a good candidate for your first autopilot. On the call, we look at whether Owner Autonomy is a fit for your business and which process is worth starting with.