The Keep-or-Kill Assessment
Keep or kill - how much of your AI spend is actually worth it?
A free 2-minute check. List your AI tools and we'll flag each one Renew, Watch, or Kill - provisionally. Connect your data to verify the real verdict, with the money at stake and a traceable confidence score.
About you
Step 1 of 2
Your Industry and Team Size give the assessment context. Your name and email are used to send you a copy of your provisional read. Your contact details are shared with the Uprovd team so the right person can follow up. * Required fields. We never share your information with third parties.
Your AI tools
Step 2 of 2 · -Add a row per tool you are billed for - Copilot, ChatGPT Enterprise, a support copilot, an agent platform. Monthly cost in USD.
Idle tool
You're paying for it, but it's barely used. Provisional: Watch / Kill.
Premium model on trivial work
A top-tier (expensive) model doing tasks a cheaper one could. Provisional: Watch.
Duplicate / overlapping tools
Two tools covering the same job, paying twice. Provisional: Watch / Kill.
You receive
Your provisional Renew / Watch / Kill read per tool, with the spend-at-risk band and a link to verify it
Uprovd team receives
Your details and provisional read so the right person can follow up within 4 business hours
Important
Every flag this page produces is provisional. It is a directional estimate from numbers you self-report - not a measured result.
Provisional spend at risk
-
-
This is a directional estimate from self-reported numbers, not a measured result. Uprovd's real verdict traces every figure to its source record with an honest confidence score; connect your data to see it. No confidence score is shown here, because self-reported inputs cannot carry one. Actual results vary depending on AI tool configuration, usage patterns, attribution methodology, and operational conditions. These flags are indicative only and do not constitute a guarantee, warranty, or representation of actual financial outcomes.
Provisional is not proof.
Connect your AI tool and cost data and Uprovd measures the verdict instead of estimating it - every figure traceable to its source record, carried with an honest confidence score.
Frequently asked questions
Is this the same as Uprovd's real verdict?
No. This is a provisional read from what you type in. The real verdict is measured from your connected AI tool and cost data, traced down to the source record, and carried with an honest confidence score. This form cannot do any of that - which is why every flag here is labelled provisional.
Is it free / do I need an account?
Free, no account required. It takes about 2 minutes and the result is emailed to you.
How is each tool flagged?
From two things you tell us: how heavily the tool is used, and how clearly you can attribute value to it. A tool with clear value and real usage flags Renew; thin usage or unattributable value flags Watch or Kill.
What does the spend-at-risk band mean?
It is a provisional band - a share of that tool's monthly cost that may not be returning. It is a directional range from self-reported inputs, not a measured figure.
What happens to my numbers?
They are emailed to you and to the Uprovd team so the right person can follow up. We never share your information with third parties.