A business saves five hours a month with AI. The owner multiplies those hours by an hourly rate and calls the result cash savings. The bank balance does not change, but the software bill arrives anyway.
The time may still be valuable. Staff could clear a backlog, improve service, or handle more work. The financial decision becomes clearer when you name that benefit accurately and include the work required to obtain it.
Your work product is a one-page investment case with separate cash and staff-capacity views. All Cedar Desk Studio figures in this chapter are hypothetical planning inputs, not measured savings or advice about Randy's finances.
Start with the decision and the alternative
Specify the workflow, the people who use it, and the decision you need to make. “Should we buy AI?” is broad. “Should we spend $300 setting up a drafting workflow and $60 a month operating it?” can be evaluated.
Describe the current alternative honestly. It might be a manual process, an existing tool, a contractor, or leaving some work undone. Include the present level of quality and service. Comparing a carefully reviewed AI workflow with an imaginary perfect manual process produces little useful information.
Choose a period for comparison, such as twelve months, and state whether the figures include taxes, financing, and inflation. The examples here use an undiscounted, pre-tax twelve-month planning view. They are not accounting statements or a valuation of the whole business.
Keep the financial terms distinct
| Term | What it means in this guide |
|---|---|
| Revenue | Sales earned under the relevant accounting approach |
| Contribution | Revenue less the explicitly listed variable costs |
| Profit | Earnings after the expenses included in the defined profit measure |
| Cash flow | Cash received and paid over a period |
| Staff capacity | Time available for other work, without necessarily reducing payroll |
| Payback | When cumulative specified benefits or cash inflows recover the specified initial investment |
The SEC's introductory guide distinguishes income statements from cash-flow statements: a business can report earnings while its cash moves differently because of timing, non-cash items, investments, and financing. See its financial statement guide.
EBITDA means earnings before interest, taxes, depreciation, and amortization. It is not a bank balance or cash flow. It does not capture every cash effect of collecting invoices, buying inventory or equipment, or repaying principal. The SEC's non-GAAP guidance defines EBITDA and discusses its presentation; do not relabel a different calculation as EBITDA.
Use the simplest measure that answers your decision. An owner deciding whether next month's bills can be paid needs a cash forecast, even if a separate profitability report looks favorable.
Count the complete workflow
For Cedar's hypothetical drafting pilot, assume:
| Input | Planning assumption |
|---|---|
| Eligible tasks per month | 100 |
| Current time per task | 8 minutes |
| Assisted time per task, including ordinary review and correction | 5 minutes |
| Additional maintenance time | 1 hour per month |
| Value assigned to staff time | $30 per hour |
| Software and usage cash cost | $60 per month |
| Initial cash setup cost | $300 |
| Initial internal training time | 4 hours |
The three-minute difference across 100 tasks releases 300 minutes, or five hours, before maintenance. At the assumed hourly value, that is $150 per month of gross capacity value. Maintenance uses one hour, valued at $30, leaving four net hours before considering software cost.
Ordinary case review is already included in the five-minute assisted time. Do not charge it again. Unusual incident handling, changes in volume, or additional oversight may still need separate allowances if they are not represented in the measured or assumed task time.
Use actual observations when available. Record task selection, failed attempts, rework, and quality. A tool that helps easy cases but creates expensive exceptions should not be judged from easy cases alone.
Show the staff-capacity view
In this explicitly valued-resource calculation, first-year gross benefit is 12 × $150 = $1,800.
Costs are $300 in setup cash, $120 in initial training time, $720 in software, and $360 in maintenance time. Total specified resource cost is $1,500. Net valued-resource benefit is $300.
Define ROI before reporting it:
ROI for this view = (specified benefit − specified cost) ÷ specified cost.
The resource ROI is $300 ÷ $1,500 = 20%. This calculation assigns a value to staff time; it does not establish $300 of extra cash or accounting profit.
Ask how the released capacity will be used. If four hours disappear into idle time and the business values only cash return, the assigned benefit may not justify the investment. If those hours remove a persistent service bottleneck, they may be useful even without a payroll reduction. Record the operational outcome you intend to verify.
Show the cash view separately
In the base scenario, Cedar does not reduce payroll, overtime, contractor spending, or another cash expense. It also has no demonstrated incremental collections. Cash benefit is therefore zero in the supplied scenario.
The project uses $300 of setup cash and $720 of software cash during the first year: $1,020 total. Its incremental first-year cash effect is negative $1,020. No cash payback occurs under these assumptions.
That result does not mean the time-saving idea is worthless. It means the proposed project consumes cash in exchange for capacity and possible service benefits. The owner can make that choice knowingly and check whether the business can afford it.
Do not use the same released hour twice. If you claim it avoids a contractor payment, do not also count that same hour as internal capacity creating additional sales unless there is separate evidence for both effects.
Calculate a genuinely different cash-saving scenario
Consider a separate hypothetical case in which the workflow actually eliminates $200 per month of contractor payments for the same scope and quality of work. Keep the $60 monthly software cost and $300 cash setup cost. Do not add the base scenario's staff-capacity benefit to this cash calculation.
Net monthly cash improvement is $140. The simple steady-state payback calculation is $300 ÷ $140, approximately 2.14 months. If cash effects occur only at month end, cumulative cash turns positive at the end of month three, not during month two.
First-year cash benefit is $2,400, specified cash cost is $1,020, and net cash benefit is $1,380. Using the same benefit-minus-cost definition, cash ROI is approximately 135.3% over twelve months.
These figures depend on the contractor payment actually disappearing and on comparable output. Internal training and maintenance still consume resources even if they do not add cash payments. Keep the resource view alongside the cash view, and verify termination costs, retained commitments, or other effects before treating a hypothetical reduction as real.
Stress the assumptions that could change the decision
At fifty tasks per month, the three-minute improvement creates 2.5 hours, valued at $75. Software and maintenance still total $90 per month in the resource view. Monthly net resource benefit becomes negative $15 before recovering the initial setup and training.
Under the base timing and cost assumptions, gross monthly capacity value is $1.50 per task. Sixty tasks cover the $90 recurring resource cost; they do not recover setup. To recover the full $1,500 first-year resource cost, the workflow needs 1,000 tasks during the year at the same per-task benefit, or about 83.3 per month on average.
Try other changes: less time saved, greater review effort, a price increase, an adoption delay, or lower usage. Do not create a “conservative” scenario by arbitrarily reducing one number while ignoring the most important risk.
If the investment lasts several years or cash timing matters materially, evaluate discounted cash flows with an appropriate rate and assumptions. Simple payback ignores value after recovery and the time value of money. An attractive short payback is not a complete investment decision.
Use AI to find omissions and explain the result
Review this investment case using the supplied assumptions and calculations.
Separate cash flows, internal time values, revenue, and contribution.
Identify missing costs, double counting, unverified benefits, and timing gaps.
Explain which assumptions could reverse the decision and propose checks.
Do not invent cost savings, sales, payroll reductions, tax treatment,
adoption rates, or approvals. Use the supplied deterministic calculations;
flag a suspected arithmetic issue for recalculation rather than guessing.
Require the underlying inputs and formulas with any conclusion. A generated recommendation that says “strong ROI” without specifying the period, benefit, and cost base is not ready for a decision.
Keep financial records in an approved environment with appropriate access. Give the model the minimum data needed for the task. A useful review often requires categories and amounts, not bank credentials or full customer identities.
Turn the investment case into a reviewable commitment
Record the owner, proposed budget, decision, expected outcomes, and date for comparing the assumptions with actual results. Keep setup costs and recurring costs visible even if the pilot stops early.
After launch, reconcile actual spending and measure the entire workflow. Verify whether staff used the released capacity and whether any claimed cash reduction actually occurred. Do not rewrite the original assumptions to make the project appear to have met them.
The investment case is complete when someone can explain why the project is worthwhile, how it affects cash, what could make it fail, and what evidence would justify continuing or stopping it.
Next: Place financial assumptions on a calendar with a thirteen-week cash forecast in Chapter 27.