AI Project Budget Planner
A month-by-month budget for an AI project: team, model and run costs, infrastructure, training and a contingency buffer, with payback from the value you expect.
- Free
- No signup
- Rates from BLS, May 2025
AI Project Budget Planner
Project definition
About 6 months and 960 hoursest. after your data and team answers.
Pick one of the three cloud options. “No cloud yet” adds a one-time setup line; the rest go in your summary.
Used in the summary.
Sets the suggested contingency buffer.
What drives the budget
Team time is usually the largest line. Three answers move it most.
Scope
A pilot proves the idea with a fraction of the hours. Each step up in scope multiplies the team time, and the months with it.
Data readiness
Messy or missing data means more hours before any model work starts. The results show how much clean data would save on your answers.
Who builds it
Full-time rates, our engineers or an agency quote change the team line directly. Compare them before you commit.
Questions
AI budget questions
Where do the team hours come from?
From HorizonLux planning figures per project type at pilot scope, multiplied by the scope you pick and adjusted for your data and team answers. They are marked “est.” until confirmed. If you have a plan or a quote, type your own total hours and they replace ours.
Where do the hourly rates come from?
Full-time rates use BLS wages for software developers (May 2025) at the 25th, 50th and 75th percentile, plus 45.9% benefits, over 2,080 hours a year. HorizonLux engineers use our monthly price over 160 hours (an estimate). For an agency, you enter the rates from your quotes.
Can we start with a small pilot?
Yes. The “Ways to spend less” section re-runs your budget at pilot scope, so you can see the difference with your own answers rather than a rule of thumb.
Why are some costs left for me to add?
Data labelling, evaluation, monitoring, documentation and tuning after launch vary too much from project to project to guess. Add your own amounts and they join the total, the month-by-month table and the payback.
How is the contingency buffer chosen?
The complexity answer suggests 10% to 25%. That is a planning rule of thumb, not a measured figure, so move the slider to what your organisation uses.
Does the budget include running costs after launch?
No. It covers the project months only. Use the monthly model and run cost as a starting point for the yearly running budget, and price it properly with the AI Cost Diagnostic.
Related tools
From budget to plan
Check the budget with people who build AI systems
Bring the budget you just made. We will go through the hours, the run cost and the risks with you, and tell you where it could be smaller.