Behaviors·Emerging

Valuemaxxing

noun

Trying to maximize the useful results of AI use, judged by outcomes such as work completed, time saved or quality improved.

Origin: Attributed to Marc Boroditsky by IBM (2026)AI Workplace MetricsEmergingUpdated Sep 2026

What did the AI actually help you do?

Valuemaxxing answers tokenmaxxing with a different measure of success: did all that activity make a worthwhile difference?

For a support team, that might mean correct answers delivered sooner, after including the time people spend checking and correcting drafts. For a writer, it might mean finishing a clearer article with less rework. These examples require someone to decide what a good result looks like before celebrating the usage numbers.

We spent more on tokens this week, but finished the migration sooner. That's valuemaxxing, as long as the fixes hold up.

Original example written for this dictionary.

Value can justify more usage

Valuemaxxing can justify using more tokens when the extra work produces a better result. It is not simply a slogan for cutting the AI bill. Cost matters alongside the usefulness and quality of the outcome.

An emerging business slogan

In a June 25, 2026 article, IBM credits Marc Boroditsky with coining the term and presents it as an alternative to token-based measures of AI adoption. That attribution comes from IBM's account. The article proposes a way to think about AI value; it does not define a standardized metric called valuemaxxing.

Source & further reading

IBM: Tokenmaxxing is dead, long live valuemaxxing