From tokenmaxxing to the invoice
A company burned millions on AI in months, built a tool to find out where it went, and is now selling it. That progression is the whole enterprise AI story compressed into one product.
Rippling shipped AI Spend Console after its own bill ran into the millions, mapping spend to individual employees, teams and roles and trying to distinguish output from slop.
Instrument-then-sell is a reliable signal
Products built from market research describe a problem someone imagined. Products built from a company's own surprise describe one that actually happened. The second kind tends to find buyers faster, because the buyers have already had the same surprise and have not yet admitted it.
The per-employee attribution is the difficult half and the valuable one. Aggregate AI spend is easy to report and impossible to act on. Spend mapped to a person, a role and an output is a performance conversation — and it makes AI usage legible to finance in a way it currently is not anywhere.
The market turned
Coverage through the summer has tracked users moving from tokenmaxxing to efficiency: buying the smallest model that clears the bar rather than the best one available. Meta is pricing directly into that shift, differentiating on cost while sitting within two points of the leaders on the benchmark.
This is what the end of a capability land-grab looks like from the buyer's side. The question stopped being what is the best model and became what is the cheapest model that does this specific job, which is the question every mature software category eventually settles on.
And the balance sheets are about to be public
OpenAI filed confidentially for an IPO following Anthropic, and both have shipped under government access limits at various points this year.
Public markets will ask the Rippling question at the scale of the whole industry. Not what can the model do, but what did the spend produce. That is a harder question than any benchmark, and nobody has had to answer it under disclosure rules before.
TechCrunch — After Rippling blew millions on AI in months, it built an employee ROI tool → · CNBC — OpenAI and Anthropic face new AI reality as users shift from tokenmaxxing to efficiency → · TechCrunch — Following Anthropic, OpenAI files confidentially for IPO →