When AI Saves an Hour but Costs You Half a Day

You might be in fewer meetings now, but you're also watching the AI more closely than you'd admit, making sure it doesn't do something you'll have to explain later. That cost is real, and it doesn't show up on any invoice.

The companies that come out ahead won't be the ones using the most AI. They'll be the ones honest about what supervising it actually costs. To be clear, we're not skeptics. AI is a huge accelerator in how we're building Leebry. We just think the math everyone's using is missing a line item.

In this issue, our Director of B2B Development, Dan Jaenicke, breaks down the real ROI of AI and why the human cost might matter more than the tokens.

There's a name for that: supervision debt. It's the review time nobody budgeted for, and like any debt, it doesn't disappear just because no one's looking at it.

That debt is still worth carrying. The person reviewing is doing two jobs: quality check, and the one who answers for it when something goes wrong. Cut them out to save money, and you cut out the value with them.

We saw the same pattern in our research. The wins were always specific, and there was always a person in the loop:

Same shape every time: one team, one task, a person signing off before anything sticks. Telling everyone to "use AI for everything" is the fastest way to watch ROI disappear.

How we think about it at Leebry

We pick narrow tasks where we can point to the return. L1 tickets are the obvious one: the average takes about five minutes to resolve, and we're working to get that under 30 seconds. If we can't see a clear return, we don't build it.

Most companies won't say this out loud: they don't actually know if it's working. Only 17% of the leaders we surveyed said even half their initiatives are delivering, and 73% have no real way to measure it. Hardly anyone can tell you what they're actually getting for the bill they're already paying.

You don't need a dashboard to start closing that gap. You need a stopwatch.

The next time your team runs a task through AI, time two things:

Add them together. That's the real number, not the one that ends up in a slide about hours saved. Do that for a week on one recurring task. If the second number is bigger than the first, you've found the line item this issue's been talking about.

That's Issue #6.

If your team's found an honest way to measure supervision debt, or if there's something else missing from the ROI math, we want to hear it. Reply and tell us.

See you in two weeks.

Leebry is a Work AI platform built by MacPaw. It connects the tools your team already uses (Confluence, Slack, Jira, Okta, HRIS and more) to give employees secure, citation-backed answers and automate everyday workflows through a single interface.