Baseline
Basic responses
- Cost / month
- $2,500
- Gross margin
- 87.5%
- Cost / user
- $2.50
Cost modeling for teams building AI products
Know your AI costs.
Protect your margin.
A new feature changes more than the product. ProfitCtl helps you model the spend, compare your options, and decide what to ship.
Try the demoMIT licensed. Runs locally. No sign-up.
01 / Try a decision
What changes when
your feature gets used?
Adjust the example. See how usage and pricing change the cost of adding research to an AI product.
Your assumptions
How many people use the product?
Monthly usage per active user.
The recurring revenue from each user.
Inspect the sample costs
Both options use $400 monthly overhead and $1.50 per user in other variable costs.
Basic responses cost $0.01 per request. Research-enabled responses cost $0.08 per request.
This example calculates gross margin. Fees, growth distributions, and full stress tests belong in your CLI scenario.
Modeled monthly revenue$20,000
With the feature
Research-enabled
- Cost / month
- $6,700
- Gross margin
- 66.5%
- Cost / user
- $6.70
The feature clears your margin target.
Research adds $4,200 per month. The modeled margin stays above 60%.
An adjustable illustration using sample costs. Your real decisions need your own rates, usage, and evidence.
02 / The workflow
From an assumption
to a decision you can defend.
The demo shows the trade-off. ProfitCtl takes it further with saved scenarios, cost sources, simulations, and explicit guardrails.
Start with what changes.
Describe usage, pricing, model calls, and infrastructure. Keep each cost tied to its source and assumptions.
Usage + rates + pricing
Put the options side by side.
Compare a feature, model, deployment, or pricing plan. See the effect on revenue, recurring margin, and cost per user.
Baseline → candidate
Set the limits before you ship.
Test margin and cost targets, including growth and tail risk. Keep the decision evidence in your review.
Margins + cost targets
Use the same scenario where your team makes decisions.
Local CLICodex skillPull request checks
03 / Keep the evidence
Inspect the assumptions
behind the answer.
A cost estimate is useful when you can explain it. ProfitCtl keeps sources, confidence, and profitability targets alongside the result.
Can I use my own costs?
Yes. Save pricing, fixed and variable costs, usage, and targets in a YAML scenario. Compare it with another option, then inspect the output.
What happens when usage grows?
Use simulations and growth scenarios to test margin and cost limits. Tail-risk checks help expose outcomes that an average alone can hide.
What is available today?
The open-source local CLI, Codex skill, and CI output are available. Hosted workspaces, saved team history, and approvals are planned.
04 / Make it yours
Bring one
real decision.
Start with a saved comparison. Then replace the assumptions with your own costs and usage.
This source-based quick start requires Git and Go. Release installation is covered in the documentation.
Terminal · Git + Go
git clone \
https://github.com/IntelIP/ProfitCtl
cd ProfitCtl
go run . compare \
examples/hybrid_steady_profit.yml \
examples/hybrid_profit.yml