01The question
How do you price software when AI writes much of the code? People still design, check and look after it.
Price the work itself, and show the working:
- 01Each feature at the size it needs
- 02The saving AI tools really bring, task by task
- 03The checking that saving demands
- 04Everything after launch
02How we test the prices
Checked against real Indian projects. Not tuned to them: our prices come from hours and rates.
- Small projects
- Real contracts mostly came in below ours: lowest prices, simpler builds.
- Medium projects
- Real contracts mostly came in a little below ours.
- Large projects
- Real contracts mostly came in above ours: hardware and licences we can't see yet.
Not checked yet: what finished projects really cost. Recorded outcomes will do that, so treat every estimate as a guide.
03What we hold to
Four rules. Every estimate follows them.
01
- Price what is built, not who asks
- A login screen costs the same for everyone.
02
- AI reads, formulas count
- AI suggests features. It never sets, edits or rounds a price.
03
- Every number has a source
- Feature, rate, where the rate came from, and the formula.
04
- Honest about how sure we are
- A range with its reasons, not one number pretending to be exact.