AUDR by Chargebee
Product Hunt fmerian
Chargebee launched AUDR, an open standard for agent run costs. It aims to show who started an AI run and what it cost, across every system it touches.
Based on reporting by Product Hunt, fmerian — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Chargebee is back with its sixth Product Hunt launch, and this time the pitch is an open standard for tracking what AI agents cost to run. The new project is called AUDR, short for Agent Usage Detail Record, and it is aimed at SaaS and AI companies trying to bill for something a lot messier than old-school software use.
The idea borrows from telecom. AUDR takes inspiration from the Call Detail Record and tries to do for agent runs what CDRs did for phone calls: create a common record that everyone can agree on. In this case, the record is a JSON schema that can be emitted and ingested by any harness, router, or billing system.
Chargebee says the standard rests on three rules. The harness mints a shared run ID. Each system owns its own fields. And if two systems disagree, conflicts are rejected rather than mashed together. Corrections are supposed to arrive as new records, not edits hidden in place.
That is a very billing-company answer to a very AI-era problem, which is probably why it makes sense. The run itself may hop across multiple systems, but the money still has to land somewhere clean. AUDR is trying to make that handoff legible.
Chargebee also frames the launch in familiar product language: billing and monetization for SaaS and AI companies, with support for a range of pricing models, from tiered and volume to usage-based and flat-fee. AUDR fits that same theme, just aimed at the part of the stack that needs the numbers to stay honest.
My take — AI-written commentary, not fact-checked reporting
Open standards are where the grown-ups go when everyone else is busy inventing a new dashboard. AUDR is smart precisely because it treats agent billing like infrastructure, not theater. If AI is going to keep selling itself as enterprise-ready, it can start by agreeing on who ran what and what it cost.
Read more about this at: Product Hunt