What would AI automation actually save your team?
Move the numbers. You get a monthly and yearly savings range — conservative to aggressive — a payback window against a real engagement cost, and a summary you can print or email. The honest version of the spreadsheet your finance team already started.
01 · Who spends the hours?
Add the roles that carry repetitive, rule-heavy work. “Automatable hours” means the part of the week a machine could credibly do most of — data entry, triage, assembly of reports, copy-paste between systems — not the whole job.
1,380 automatable hours a year at $38/hour — worth $52,440 if none of it is recovered.
02 · The engagement to pay back
A typical mid-market workflow automation engagement — audit plus the first production pipelines — lands in this band. Replace it with a real quote if you have one; payback below is measured against both ends.
03 · The numbers
Across 3 people and 30 weekly hours of automatable work at a blended $38/hour. Conservative assumes 30% of those hours come back with the messiest workflows and the highest tooling costs; aggressive assumes 60% — still below our published logistics case.
Hours recovered / month
35 – 69
of 115 automatable
Net / year, steady state
$14,904 – $30,636
after tooling costs
Net, first year
$13,041 – $26,807
ramp included, not month one
Payback window
12 months – past 36 months
vs $25,000–$75,000 engagement
| Role | People | Automatable h/yr | Net/yr, conservative | Net/yr, aggressive |
|---|---|---|---|---|
| Operations & admin | 3 | 1,380 | $14,904 | $30,636 |
From estimate to measurement
Every number above rests on your estimate of where the hours go. Our workflow automation engagement replaces those estimates with measurements — a two-week audit of how your team actually spends its time, then production pipelines with human checkpoints where judgment matters. The savings dashboard we leave behind counts the hours, the error rate, and the spend continuously, so the payback line above becomes a tracked number instead of a projection.
Inputs stay in your browser · no email gate, no account · nothing is sent unless you ask us to email the estimate · the arithmetic runs on this page
Email me the full report
Leave your address and we send the full savings estimate to your inbox — a human presses send, so it arrives from hello@thethirdwaycompany.com, usually within one business day. The report on this page stays exactly as it is either way.
AI automation savings estimate
Generated 16 October 2026 at thethirdwaycompany.com/tools/ai-savings-calculator — a rough model, not a quote.
| Role | People | Automatable h/wk/person | Loaded $/hour |
|---|---|---|---|
| Operations & admin | 3 | 10 | $38 |
| Conservative | Aggressive | |
|---|---|---|
| Hours recovered / month | 35 | 69 |
| Net savings / month (steady state) | $1,242 | $2,553 |
| Net savings / year (steady state) | $14,904 | $30,636 |
| Net savings, first year (ramp included) | $13,041 | $26,807 |
Payback against a $25,000–$75,000 engagement: 12 months (aggressive savings vs the low end) to past 36 months (conservative savings vs the high end).
Assumptions: 46 paid working weeks a year; conservative recovers 30% of automatable hours at $2/hour tooling; aggressive 60% at $1/hour; savings ramp over 3 months then run at full rate; net = value of recovered hours minus tooling. The hours are your estimate, not a measurement — our workflow audit replaces them with measurements: thethirdwaycompany.com/services/ai-workflows.
How the math works
Each role contributes people × automatable hours/week × 46 weeks of automatable hours a year. The two scenarios then price what actually comes back: the share of those hours recovered, minus what the AI tooling itself costs to run. Savings do not start on day one — the audit and the first builds come first — so the model ramps in over 3 months before running at full rate, and payback counts cumulative net savings against the engagement band, month by month, for up to 36 months.
| Scenario | Share recovered | Tooling / recovered hour | When it is true |
|---|---|---|---|
| Conservative | 30% | $2 | Exceptions still reviewed by people, adoption slower than hoped, workflows carry edge cases. Tooling priced high. |
| Aggressive | 60% | $1 | Structured, rule-heavy work with clean data access — the good end of the range, still below our published logistics case. |
Both shares are our editorial estimates, and both sit deliberately below the two-thirds of automatable hours our published logistics case actually recovered — that engagement is why the aggressive end exists, and why it is a case study rather than a default. The same reasoning, priced from the other side, is in our cost-model post: the unit cost of AI is not the cost — the outcome is.
What this number is — and is not
It is a conversation starter with the arithmetic shown. The hours are your estimate of what counts as automatable; people argue about that number more than about any other input here, and the argument is the useful part. When we run a workflow audit we replace the estimate with measurement — shadowing, tool instrumentation, and interviews — because teams consistently misjudge where their own hours go.
It is not a headcount plan. The tool prices hours, not jobs: automation shifts work — someone still reviews exceptions, and the conservative share of 30% already prices that in. In practice the outcome is the same people doing higher-value work, which is the position we argue for openly on the thesis page: automate the repetitive, keep the judgment, and staff accordingly.
And it is not a steady state on day one. The ramp is real — first workflow live in weeks, full run-rate a quarter in — which is why the first-year number is lower than a twelfth of the annual one, and why payback measured honestly usually lands in months, not weeks.
Turn the estimate into a tracked number
Every projection on this page becomes honest the moment measurement replaces assumption. That is the first deliverable of our AI workflow automation engagement: a ranked map of where the hours actually go, then production pipelines for the workflows with the best savings-to-risk ratio — agents for the volume, humans for the exceptions.
The second deliverable keeps us honest: the savings dashboard. Hours saved, error rates, cycle times, spend — counted continuously after deployment, so if a workflow is not paying for itself we fix or retire it. Bring this estimate to the first call and the audit has a number to beat. Or start with a conversation — the calculator is free either way, and nothing you typed into it leaves your browser unless you ask us to email the estimate.
One email a month on practical AI adoption
What we ship for clients, what the tools can honestly do this month, and one thing worth trying in your own team — written by us, no hype, unsubscribe anytime.