Pick the task, tell us roughly how much time it takes, and get an honest estimate — no call required to see the number.
This is a free estimator for one question: if a repetitive task in your business were handed to an AI system, how much time and money would come back? You pick the kind of work, say how many hours a week it takes and how many people do it, add a rough hourly cost, and you get three figures — hours reclaimed per month, savings per month, and savings per year.
It is not a quote, and it is not a promise about your business. It is a way to find out whether a workflow is even worth scoping before anyone spends engineering time on it. Most of the value is in the comparison: run two or three tasks through it and the one worth automating first is usually obvious, often not the one that felt most annoying.
The arithmetic is deliberately simple, and it is the same four steps for every task type:
Worked through with ordinary numbers: 3 people spending 8 hours a week each on the same task, at $30 an hour, is 104 hours a month going into it. At 60% that is 62 hours reclaimed a month, worth $1,871 a month or $22,447 a year in labour cost. Roughly a third of a full-time role, recovered from one workflow.
Almost nothing goes fully hands-off. Exceptions get escalated, edge cases need judgment, and anything customer-facing or regulated keeps a human in the loop — which is a feature of a system you can trust, not a failure of the automation. So the calculator assumes automation takes over about 60% of the time currently spent and leaves the rest with your team.
One flat rate is applied to every task type, and that is a deliberate choice. Varying it by task — 75% for support, 55% for reporting — would look more rigorous, but we have no dataset that would make those numbers true. False precision is worse than a stated assumption you can argue with.
The hours figure is the one to trust most, because it depends only on your own inputs. The money figure depends on what an hour of that time is worth, which is where estimates usually go wrong. Use fully loaded cost — salary plus employer taxes, benefits, tooling and overhead — rather than take-home pay. Annual salary ÷ ~2,080 hours, plus 25–40%, gets close enough.
It is also worth being clear with yourself about what the reclaimed hours are for. If they go back into sales calls, product work or customers, the saving is real economic value. If the team is simply under water and the hours buy breathing room, the value is real too, but it shows up as retention and fewer mistakes rather than as a line in the accounts. Both are good reasons to automate. They are not the same business case.
The estimator covers the five categories we are asked about most, and each one comes with a different shape of automation:
If a task is rare, or genuinely creative, the estimate will come out small. That is the tool working correctly. Automate the high-volume, low-judgment steps first and revisit the rest once the first one is running.
Treat the output as gross labour-time value. A real business case subtracts build and run costs from it — which is exactly the arithmetic that turns this estimate into ROI, and the part we would rather do with you than guess at here.
If the figure is large enough to be interesting, the next step is not buying a chatbot. It is mapping the workflow properly: what triggers it, which systems it touches, where a human approves, and what happens when something fails. That map is what separates automation that survives contact with a real week from a demo.
We build both ends of that. AI automation and generative AI workflows wired into the products you already run, and agentic AI development for systems that plan multi-step work, call your tools and act with guardrails rather than answering in a chat window. If you would rather read before you talk to anyone, the free playbooks are the longer version of how we work.
Not quite. ROI is savings divided by what you spend to get them. This tool only estimates the first half — the labour time and cost currently going into one repetitive task, and the share of it automation could reasonably take over. To turn that into ROI, divide the annual figure by what the build plus a year of running it would cost. We'll give you that second number on a scoping call; the calculator deliberately doesn't guess at it.
It is directional, and it is only as honest as the hours you type in. Three things move it: whether the hours reflect real time-on-task rather than an impression, whether the hourly rate is fully loaded, and whether 60% automation is realistic for that particular workflow. Use it to rank which process to automate first, not to forecast a line in next year's budget.
Use the fully loaded cost of the people doing the work — salary plus employer taxes, benefits, software and overhead — not their take-home pay. A quick approximation is annual salary divided by about 2,080 working hours, plus 25–40% for overhead. If you are unsure, enter the lower figure: an estimate that survives a conservative rate is the one worth acting on.
Because almost no real workflow goes fully hands-off. Exceptions get escalated, edge cases need judgment, and anything customer-facing or regulated needs a human in the loop. The calculator applies one flat assumption — automation absorbs about 60% of the time spent — to every task type. One rate rather than a different figure per task, because we have no dataset that would make per-task percentages anything other than false precision.
No. The hours and the savings figures appear as soon as you enter your numbers, and you can change the inputs and recalculate as many times as you like without a form. The name and email field further down is for something else — the specific recommendation for how we would automate that task, and a copy of the estimate in your inbox.
No, and that is the main thing to hold in mind when you read the number. It is gross labour-time value only. A real business case subtracts the one-time build, model and API spend, tool seats, maintenance, and the review time that stays human. High-volume tasks at a decent hourly rate usually clear those costs comfortably; low-volume ones often don't, which is worth knowing before you commission anything.
Yes, one at a time. Run it for each workflow separately and add the monthly figures together — that stacked total is usually the honest business case, because no single task is normally big enough to justify an automation programme on its own. The calculator works in months of 4.33 weeks, so the figures from separate runs are directly comparable.