What a managed role actually costs — and how to work out whether it's worth it

There's a bill your business pays every month that never arrives as a bill.

It's in the Friday afternoon someone spends working through overdue invoices. It's in the quote that took two days to go out, by which point the customer had already accepted one that took two hours. It's in the enquiry that landed at 6pm on a Thursday and got answered on Monday morning, to someone who had stopped looking.

Nobody itemises that. There's no line in your accounts called "work that didn't happen on time". Which is precisely why it never gets fixed — you can't cut a cost you can't see, and you certainly can't compare it to a price.

So let's put both on the table. Here's what a managed role costs, what you actually get for it, and how to work out whether the numbers favour you. Including when they don't.

What you're actually buying

A managed role is a defined scope of work that an AI agent takes over and runs continuously, and that we monitor, maintain and report on every month.

Some concrete ones: a receivables clerk that works through your overdue invoices every week and tells you at month end how much came back. A quotation specialist that takes an enquiry, asks the questions you'd ask, and returns a priced estimate calculated from your own price list. A booking coordinator that handles requests, confirmations, reminders and reschedules around the clock.

If the word "agent" is doing a lot of work in that sentence, start here: What is an AI agent, actually? · All managed roles we run

Now the part most people selling this leave out.

It is not a person. It doesn't do everything someone in that job would do. It covers the scope we agree at the start — written down, testable, reported on. If your receivables clerk would also notice that a particular customer always pays late and mention it in the corridor, the agent won't do that unless we scoped it in.

It is not a tool you operate. There's no dashboard for your team to learn and abandon. You don't log in.

It is not a one-time build. Software you commission and never touch again is software that quietly stops matching your business. The monthly fee is somebody keeping it working as things change.

That's the shape of it. Now the price.

What it costs

Two things worth explaining, because the shape of this is deliberate.

Onboarding is one-time and per account, not per role. It buys the part nobody else can do for you: connecting an agent to your systems, your price list, your calendar, your way of doing things. It scales with how many roles and how many integrations are involved — a single role wired to one inbox sits at the bottom of that range, three roles across several systems at the top. Once that connection exists, adding a second role is cheaper, which is why the second one costs less than the first.

The monthly fee is the actual product. It isn't a maintenance charge to stop things breaking. It's somebody running the thing: watching what it does, correcting it when your business changes, and reporting what happened. We used to price this the other way round — a large build fee and a small monthly one — and it was wrong. It charged the most for the part that's becoming commodity and the least for the part that actually matters.

What we won't do is compare that to a salary. It's tempting, and every competitor does it, but it sets a false expectation: anchor a partial-role product to a whole person's pay and you'll rightly expect whole-person judgement. A managed role is narrower than a hire and should be priced and understood as such.

How to work out whether it's worth it

Three questions. The third is the one that matters most and gets asked least.

1. How many hours does this actually take? Most owners underestimate by about half — not out of carelessness, but because the visible part is the smallest part. Chasing an invoice is four minutes of typing and twenty minutes of working out who to chase, finding the reference, and deciding how firmly to word it. Count the interruption, the context-switch, and the second attempt. Then count the times per week.

2. What does it cost when it doesn't happen? This is where the real money usually is, and it's the number nobody has. Invoices paid at 45 days instead of 30 tie up cash you already earned. Enquiries answered on Monday convert worse than enquiries answered in five minutes. Quotes that arrive third rarely win. No-shows that a reminder would have prevented are pure loss, because the slot can't be resold retroactively. You won't have exact figures for these. Estimates are fine — you know your business better than any benchmark does.

3. What would you do with the hours instead? If the honest answer is "nothing especially valuable", then automation may not be worth it, and you should keep your money. Time returned only becomes money if it goes somewhere useful — selling, building, or simply not working on Sunday. There's no shame in the third one, but be honest about which it is.

Work it out on your own numbers

We built a calculator that does this arithmetic. Pick a role, put in your figures, and it tells you what comes back per month, what it costs, and how long it takes to pay for itself.

Every assumption is visible and editable — how much you think consistent chasing would improve your collection times, what share of quoting is genuinely automatable. Change them if ours look wrong for you. There are no hidden multipliers, and nothing is rounded in our favour.

It will also tell you when the answer is no. That's not modesty; it's the more common outcome for smaller operations, and we'd rather you found out in ninety seconds than after signing something.

No email required. The result is on screen. Work out whether a managed role pays for your business →

When you shouldn't do this

An agent is the wrong answer more often than the industry admits. Specifically:

• When the task is rare. Something that happens twice a month doesn't justify integration work, whatever the hourly maths says.
• When the process changes constantly. Agents are good at work with a stable shape. If the way you quote is different every time because every job is genuinely bespoke, you need a better template, not automation.
• When the volume is small. Below a certain threshold the arithmetic simply doesn't work, and no amount of framing fixes that. Grow first.
• When the work needs judgement more than consistency. Negotiating, handling an upset customer, deciding whether to make an exception — those are people-jobs and they're staying people-jobs. What an agent takes over is the repetitive shell around them.

If two or more of those describe your situation, fix the process manually first. That's cheaper, faster, and sometimes it turns out to be enough.

Frequently asked questions

If you're not sure

The calculator handles the arithmetic. If you'd rather have someone look at your actual business, we do that free: send your website and a description of the repetitive work, and we'll send back three to five things worth automating, ranked, plus what to leave alone. Within 48 hours.

Two ways to get a straight answer: Get a free AI Automation Opportunity Report → · Is it safe to give AI your customer data?

Or book a free 30-minute audit and we'll talk it through. If an agent isn't the right fix, we'll say so on the call — that's a cheaper conversation for both of us than the alternative.