World Model Readiness The costs the pitch left out

Module · The costs the pitch left out

The AI Budget Reality Check

The vendor quote is the smallest number you will pay. This module checks whether your budget accounts for the work that surrounds every AI system: cleaning the data, governing it, integrating it, keeping it running, and retraining it as reality drifts. It also asks the uncomfortable questions: is there contingency, and who signs off when the real number lands?

Question 1 of 5 · Data work is budgeted

Does your budget include the cost of getting your data ready for AI?

Most AI budgets assume the data is usable. It rarely is. Cleaning, labelling, connecting, and maintaining data is often the largest line item, and it is the one vendors are happiest to leave off the quote.

Question 2 of 5 · Full TCO is known

Have you costed the whole life of the system, not just the licence and the build?

Integration, governance, monitoring, support, and retraining run for as long as the system lives. A budget that stops at go-live is a down payment mistaken for a purchase price.

Question 3 of 5 · Running costs are funded

Is there ongoing budget to maintain and retrain the system after it launches?

AI systems decay: the world drifts away from the data they were trained on. Without a standing line for monitoring and retraining, performance quietly erodes until someone notices the results stopped being useful.

Question 4 of 5 · Contingency exists

Is there a contingency reserve for the parts of this that will cost more than planned?

AI projects surprise on the downside more than the upside. A budget with no reserve is a plan that assumes the first estimate, made with the least information, was correct.

Question 5 of 5 · Sign-off is honest

Does the person who signs off the budget see the real total, not the optimistic one?

Numbers get trimmed on the way up to make the case approvable. If the decision-maker approves a figure everyone privately knows is too low, the overrun is booked before the project starts.

For the statistics · one click each

Three questions for the public picture

These do not affect your score. They feed the anonymised, aggregated statistics; groups under 8 respondents are never shown.

How was your AI budget set?

From vendor quotes
Internal estimate
Full TCO model
No formal budget
Not sure

How much contingency does your AI budget carry?

None
Under 10 percent
10 to 25 percent
Over 25 percent
Prefer not to say

Have past AI or IT projects come in over budget?

Rarely
Sometimes
Usually
Almost always
No track record yet

Your context

Used to calibrate the report. Company size and sector remain in the anonymized dataset; your email does not.