Define one task.
Start with a specific person and a repeatable activity. Set a monthly volume and budget. Presets help explore the interface; they do not represent observed behavior.
Everyday AI access.
What does useful AI cost when you count the conditions around it?
EXP / 001 / THE QUESTION
Access begins before
the first answer.
Subscription fees are only one condition. A person also needs a compatible device, enough data, a workable connection, and an answer that helps them finish a task.
This browser prototype makes the financial assumptions visible. Change the conditions, compare two delivery models, and take a scenario into a more complete access review.
MODEL / THE COST OF PARTICIPATION
AFFORDABLE ≠ USEFULREVIEW BOTH
WORKING BROWSER TOOLNO ACCOUNT REQUIREDINPUTS STAY IN YOUR BROWSER
Interactive prototype / 01Local calculations +
Everyday AI Access
What does access actually cost?
Change the assumptions. See what a monthly budget can support, and where the costs accumulate.
Illustrative scenarios, not provider prices or measured performance. All values stay in your browser until you choose to download them.
03 / Read the trade-offsScenario model
Enable JavaScript to compare scenarios locally. The equations and assumptions below are available without it.
Complete the assumptions
Enter valid values in the marked fields to compare this scenario. Results are paused until then.
JSON / assumptions + calculations
04 / Equations & assumptions
Follow the calculation.
- Monthly data
- Tasks × MB per task ÷ 1,000 = GB
- Pay-per-task total
- Device overhead + (tasks × task price) + data cost
- Subscription total
- Device overhead + subscription price + data cost
- Affordable tasks
- Floor of (budget − fixed costs) ÷ variable cost per task
- Transfer time
- Tasks × MB per task × 8 ÷ Mbps = seconds
What this can tell you.
A comparison of financial and data-transfer assumptions. Both delivery options are assumed to attempt the same tasks with the same data use.
- Affordability does not measure usefulness, quality, reliability, privacy, or accessibility.
- Transfer time excludes inference, latency, retries, network overhead, and human review. It is an idealized estimate, not a speed test.
- A zero-speed connection cannot transfer a positive amount of data, even if the scenario is affordable.
- Task capacity is zero when fixed costs exceed budget. With zero variable cost and affordable fixed costs, the cost model sets no task limit.
- Tax, currency changes, subscription task caps, and price tiers are excluded. Add applicable fixed charges to overhead; check real terms before relying on a scenario.
02 / METHOD & BOUNDARIES
Make the assumptions
inspectable.
A cost model answers a narrow question. Whether a tool is accurate, understandable, available, and worth using requires a separate evaluation.
Account for participation.
Compare service cost plus data cost plus a monthly device allocation. Use your own prices and assumptions. Record taxes, training time, language, and regional restrictions separately.
Check usefulness.
Agree what a successful task looks like before testing a tool. Review accuracy, clarity, correction effort, and recovery when a task fails. A lower price does not establish a better outcome.
Make a bounded decision.
Export the scenario, complete the worksheet, and identify the evidence needed next. Treat estimates as provisional; update them when actual prices and observations are available.
WHAT THIS MODEL LEAVES OPEN
It does not call an AI model, measure inference speed, establish answer quality, or verify a service’s regional availability. Transfer time is a simplified network estimate. Subscription capacity assumes no service quota; enter actual restrictions in the worksheet.
03 / DECISION LOG
A record that can evolve.
The prototype is available. Participant research and service comparisons remain future work.
Make the cost model usable.
Added editable scenario inputs, two delivery models, visible formulas, and a local scenario export. Released a companion worksheet and worked example.
Does the model reflect a real task?
Choose one activity with intended users. Replace sample assumptions with sourced prices, data measurements, and a clear definition of a useful result.
Which delivery conditions matter most?
Investigate failure, language, support, and regional availability alongside cost. No participant study or measured impact is reported here.