The cost is recorded. The context is scattered.
We built ai-cur to close that gap: one record from the vendor's bill to the work it helped, for every function, with the evidence behind each connection.
The bill and the result live in different buildings.
AI spend arrives as seats, tokens, credits and platform charges from a dozen vendors. The work it helps is recorded where work has always been recorded: in the repository, the ticket queue, the pipeline. Finance sees one half, each team sees a slice of the other, and the question every leader is now asked, what did we get for it, has no system behind it.
ai-cur is that system. It is deliberately modest about what it claims. It does not say a tool did the work. It puts the spend and the work side by side, says how strong the link is, and leaves the verdict to the people accountable for it.
Our product views retain source coverage, cost basis and allocation rules so the reader can inspect how a result was formed.
Read the changelog and desktop client guide.
Six principles, each one visible in the product.
Evidence before claims
Every figure shows the cost and the count it came from, the sources it had, and how strong the link is.
Never the content
Amounts, people, dates and states. No prompt, file, diff or ticket body, from any connector.
Visibility, not watching
People see themselves, leads see their team, leaders see groups above a minimum size.
A record that holds up
An append-only ledger, corrections beside the original, and no total that mixes currencies.
Clear commercial terms
Contact us to discuss the scope, availability and commercial terms for your team.
Yours to run
On your servers, in Databricks or in ServiceNow, and hosted by us once sign-in opens, in a FOCUS-aligned cost format.
Built by Harvto LLC, in California.
ai-cur is a product of Harvto LLC, a California limited liability company. We are a small team that ships most weeks and answers its own email.
See what your AI spend returned this month.
Coming soon. Contact us for availability.