Revenium 2.19.0 is about attribution at the level where work actually happens. Coding assistants, agents, and automated jobs generate spend all day, and until now the picture stopped at the tool. This release pushes it down to the skill, the task, and the individual job, so a number on a dashboard comes with the story behind it. It also opens the door to Claude Enterprise customers with a connection path that takes almost no setup.
Here are the biggest things in this release.
1) Skills and Tools show what your Claude Code usage actually costs
A new page under AI Assistants shows which Claude Code skills your teams use, what each one costs, and how adoption is trending over time.
Coding-assistant spend has been visible for a while. What was missing was the layer underneath it. A team could see that Claude Code cost a certain amount last month without knowing which skills drove that spend, which ones caught on, and which ones were built and then quietly ignored.
What this unlocks:
- See which skills are earning their keep and which are expensive or unused, with per-skill cost, usage, and trend in one place
- Attribute coding-assistant spend to the specific skills driving it rather than to the assistant as a whole
- Guide enablement by spotting the skills power users rely on and rolling them out to everyone else
2) Claude Enterprise teams can now connect in minutes
Customers on Claude Enterprise can now connect to Revenium using a tightly scoped, analytics-only API key. Revenium detects the plan automatically and picks the right connection path, so there is no manual configuration required to get usage flowing.
What this unlocks:
- Claude Enterprise accounts can connect with an analytics-only, least-privilege API key
- Automatic plan detection selects the correct connection path with no manual setup
- Claude Code cost and skill visibility for Enterprise-plan organizations
3) A redesigned Tasks tab that shows where volume and cost concentrate
The Tasks tab under Performance has been rebuilt around a simple question, “Where is your AI doing the most work, and what is that work costing you?”.
The tab now brings together a volume overview, trends, a concentration breakdown, and a per-task-type performance table, with a drill-in available for any task type. Assembling that picture by hand across separate views is no longer part of the job.
What this unlocks:
- See where task volume and cost concentrate across everything your AI does
- Spot underperforming task types at a glance and drill straight into them
- Compare task types on one table instead of stitching views together
4) Failed and canceled jobs now explain themselves
When an AI job fails or is canceled, Revenium now shows the reason right next to the job. Spend on failed runs becomes explainable instead of an unaccounted line in the total.
Job outcomes, amendments, and history are reported through the public job-tracking API, from any language or integration point. The reason travels with the job rather than living in a log somewhere else.
What this unlocks:
- No more mystery failures, because the reason the agent or SDK reported sits right there with the job
- Spend on failed runs becomes explainable and actionable
- Outcome reporting from any agent stack through the public job-tracking API
5) Insights notifications go where your team already looks
On the Insights page, you can now pick which webhook channels receive notifications when an analysis run completes, partially completes, or fails, or when someone leaves feedback on a recommendation.
No polling, and no separate dashboard to remember to check. Findings arrive in the same place as the rest of your operational signal.
What this unlocks:
- Route AI optimization findings into the channels your team already watches
- Stop polling for results, because runs announce themselves
- Notifications on run completion, partial completion, failure, and recommendation feedback
6) Integration sync health for GitHub and GitLab
GitHub and GitLab connections now show their sync status on the Integrations page, with a sync-log view for the full history. Missing permissions are caught at connect time rather than surfacing weeks later as a gap in the data.
Silent integration failures are one of the most expensive problems in cost tooling, because the numbers still look fine while quietly going stale.
What this unlocks:
- A degraded or failing sync is visible on the card with its cause, instead of hiding behind a green badge
- Permission problems surface when you connect, not weeks later as missing data
- A sync-log view for connection history
Also in this release
Alongside the headline features, 2.19.0 rolls up a set of fixes and smaller improvements.
- Search added to the GitHub PRs and GitLab MRs member tables
- Small non-zero AI spend amounts now display as "< $0.01" instead of $0.00
- Read-only accounts can now manage their own password and timezone
- The Azure provider view no longer includes non-AI costs
For the full technical details on everything that shipped, the changelog has the complete rundown.
Try it out
Revenium 2.19.0 is live in production.
If you are already using Revenium, start with Skills and Tools, the redesigned Tasks tab, and job failure reasons. If your team is on Claude Enterprise, the new connection path is the fastest way to get Claude Code spend into Revenium. Together, they connect AI spend to the specific work behind it, which is where most cost decisions actually get made.
For the full list of improvements, fixes, and technical details, read the public changelog at docs.revenium.io/get-started/changelog.
