Free calculator · business case in 60 seconds · updated Sep 2026
AI ROI calculator for automation and agents
This AI ROI calculator builds the business case the way a CFO reads it. Hours saved per week, times people, times a fully-loaded hourly rate, times 52 gives the annual labour value. Then it subtracts what most calculators skip: the one-time build cost, the monthly run cost for inference, hosting and monitoring, and a ramp-up period before the agent delivers full benefit. You get the payback month, 12-month and 3-year ROI, and a cumulative cash curve you can paste into the approval deck.
Table view
| Period | Cumulative net | Break-even |
|---|---|---|
| M1 | -33.5K | 0 |
| -29.7K | 0 | |
| -23.4K | 0 | |
| -17.2K | 0 | |
| -10.9K | 0 | |
| M6 | -4.7K | 0 |
| 1.5K | 0 | |
| 7.8K | 0 | |
| 14K | 0 | |
| 20K | 0 | |
| 26K | 0 | |
| M12 | 33K | 0 |
| 39K | 0 | |
| 45K | 0 | |
| 51K | 0 | |
| 58K | 0 | |
| 64K | 0 | |
| M18 | 70K | 0 |
| 76K | 0 | |
| 83K | 0 | |
| 89K | 0 | |
| 95K | 0 | |
| 101K | 0 | |
| M24 | 108K | 0 |
| 114K | 0 | |
| 120K | 0 | |
| 126K | 0 | |
| 133K | 0 | |
| 139K | 0 | |
| M30 | 145K | 0 |
| 151K | 0 | |
| 158K | 0 | |
| 164K | 0 | |
| 170K | 0 | |
| 176K | 0 | |
| M36 | 182K | 0 |
- Fully-loaded hourly rate
- $57 (salary / 2,080 x 1.4)
- Monthly benefit at full ramp
- $7,140
- Year-one total cost
- $45,800 (build + 12 months run)
- 12-month ROI
- 71%
- Benchmark
- Microsoft example: 45 min/day for a $75K worker = $5,625/user/year
The agent pays back in month 7 and returns 271% over three years on $85,680 of annual labour value.
Assumptions and sources (7)
| Constant | Value | Basis |
|---|---|---|
| Fully-loaded multiplier | 1.4x salary (published range 1.3 to 1.5) | sourcedAI Makers, AI automation ROI formula |
| Working hours per year | 2,080 | sourcedStandard 40 h x 52 weeks (same source) |
| Savings formula | hours x people x share x loaded rate x 52 | sourcedSolvSpot ROI estimation method |
| Ramp-up | linear from 0 to full benefit over the ramp months | proxySolvSpot calculator input |
| Default monthly run cost | $900 | proxyInside the $200 to $2,500 published run-rate band (SoftTeco) |
| Default build cost | $35,000 | proxyMid-market band $25K to $120K (Decipher Zone / Musketeers) |
| Benchmark line | $5,625 per user per year | estimateEPC Group reading of a Microsoft Copilot example (third-party estimate) |
Key takeaways
- Labour value is hours x people x loaded rate x 52. Everything else is subtraction.
- Run cost and ramp-up are the two lines most ROI calculators skip. Both are here.
- 57% of enterprises say AI ROI fails to outpace spend. The usual cause is a pilot that never scaled.
- Six to fourteen months is a realistic payback for a mid-market agent that reaches production.
How to calculate ROI of AI automation
Start with the workflow, not the technology. Count the hours per week each person spends on it. Multiply by people, by the share the agent will take over, and by a fully-loaded hourly rate: salary divided by 2,080, times about 1.4 for benefits and overhead. Times 52 is the annual labour value.
Then subtract. The one-time build cost. Twelve months of run cost for inference, hosting and monitoring. A ramp period where the benefit climbs from zero. ROI is net benefit over total cost. This is the same arithmetic Microsoft publishes for Copilot Studio agents, applied to a custom build. TheoSym runs it with you on the AI consulting call before any prompt is written.
- Annual value = hours x people x share x loaded rate x 52
- Net month = benefit(t) minus run cost
- Payback = first month cumulative cash is positive
What is a good ROI for an AI agent?
A mid-market build of $25K to $120K that removes several hours a week from a team of five to ten typically pays back in six to fourteen months, if it reaches production. Vendor claims of 150% to 400% assume adoption and governance. Poorly governed rollouts land near zero. Domino Data Lab found 57% of enterprises say AI ROI still fails to outpace spend, unchanged since 2025 (Domino/BARC 2026).
The difference between the two outcomes is rarely the model. It is whether the workflow was redesigned around the agent and whether an eval suite proves it works before people rely on it. The AI Factory workproofs show what that looks like when it ships.
AI ROI calculator vs hours saved calculator
An hours-saved calculator stops at labour value. An ROI calculator keeps going: it charges the build, charges the run-rate every month, and delays the benefit while the team adopts the tool. The first number is always bigger. The second is the one finance approves.
Run cost is the line to watch. Small and mid-sized agents cost $200 to $2,500 a month to run; enterprise agents far more. Buyers are now advised to model a 12-month run-rate beside every build quote. TheoSym's build packages include that estimate and the eval suite the number depends on.
When do you need an ROI model before building an agent?
Whenever someone has to sign. A written ROI model with sourced assumptions turns a pilot request into a capital decision with a payback month. It also tells you what not to build: if the automation share has to exceed 90% to pay back, the workflow is wrong for an agent today.
For regulated workflows the model has one more line: the cost of a wrong answer. Credit, underwriting and reporting agents route to QGI, where the decision layer is deterministic and the exposure model is different. Press and publications about both companies are in the press center.
AI ROI calculator: questions people ask
How does this AI ROI calculator work?+
Take hours saved per week, multiply by people affected, by the fully-loaded hourly rate (salary divided by 2,080, times about 1.4), and by 52 weeks. Subtract the one-time build cost and twelve months of run cost, then divide by total cost. The calculator adds ramp-up so the first months are not overstated.
What is a realistic payback period for an AI agent?+
Mid-market builds of $25K to $120K that remove several hours per week from a team of five to ten typically pay back in six to fourteen months, provided the agent actually reaches production. Published vendor ROI claims of 150% to 400% assume strong adoption and governance; poorly governed rollouts land near zero.
Why do 57% of enterprises say AI ROI fails to outpace spend?+
Domino Data Lab's 2026 survey of 639 senior AI leaders found ROI stalled for two years even as production capability improved. The common causes are pilots that never scale, no workflow redesign, and unbudgeted run costs. Modelling run-rate and evals up front is what this calculator is for.
Should run costs be included in an AI ROI calculation?+
Yes. LLM inference, hosting and monitoring typically add $200 to $2,500 per month for small and mid-sized agents and far more at enterprise scale. Buyers are advised to model a 12-month run-rate beside every build quote. Leaving it out is the most common reason business cases look better than results.
What does the ramp-up input do?+
It delays the benefit. With a two-month ramp the agent delivers a third of its full value in month one, two thirds in month two and the full amount from month three, while run cost is charged from day one. That is closer to how adoption works than assuming full benefit at launch.
Next step
Turn the estimate into a scoped plan
Send the calculator result. You get one reply from a human within one business day, with an eval plan for the workflow and a real scope.
Prefer to talk? +1 657-888-0688 or contact@theosym.com