The Meter Doesn't Care Which AI You Use
An AI-first read on Atlassian's usage-based pricing.
Your customers, your admins, and probably your inbox found out about usage-based pricing at the same time you did. The questions are already moving faster than the answers.
So here is the calm version.
Something changed today, and it is worth understanding before it is worth reacting to. The pricing is new. The thing it measures is not.
The shape is simpler than the noise around it
Your seats did not go away. A usage layer went on top of them, for the parts of the platform that scale with how much work you ask them to do rather than how many people you employ. Most paid plans include real allowances before any of that is billable, and Atlassian has published a runway before billing begins.
The meters, the numbers, and the dates all live on Atlassian's usage-based pricing page, and that is where to send anyone who wants specifics. Read the page for the figures. Read this for what they mean.
The meter worth understanding first is the one on your AI
Rovo credits now cover two things: the AI features you already know, and something quieter, the enriched calls that pull context across your tools.
Here is the distinction that matters. Looking something up or changing it inside a single product stays free. The work that draws on the meter is the enriched work: reaching across Jira, Confluence, and the systems connected to them, reconciling what they say, ranking it, and filtering it by who is allowed to see what.
You are not paying for the answer. You are paying for the reach behind it.
The meter is on the reach, not the tool
This is where the AI That Works lens earns its keep.
Regular readers know the refrain. AI is only as good as the context underneath it. Take that as given. What is new is that the context now has a meter on it.
And that meter does not care which tool you use. Reach your Atlassian context through Rovo, through the MCP server, through the command line, or through an outside assistant, and the enriched call underneath is the same call. The front end is a choice. The reach underneath is the cost.
Which reframes the question everyone is about to ask. It is not “which AI tool is cheapest.” It is “where does our work pull context, and how often.”
You can change the face of your AI whenever you like. You are still drawing on the same graph.
The AI meter is not the only one
It gets the headlines. The others are worth a look so nothing surprises you.
Automation is the one more teams will feel, because it now counts the work inside each rule rather than just the run, so a busy rule counts for more than it used to. Assets is reaching more of the platform. Bitbucket's meters are moving into one place. And in Customer Service Management, an AI agent is counted only when it resolves a request from end to end, with no human behind it.
Each of those has its own details, and its own page. Go find your own footprint before the meter finds it for you.
What is worth doing while it is still free
Turn on the usage view in Atlassian Administration and read it. Watch where consumption gathers in your own environment, because your pattern is the only one that matters, and it will not look like anyone else's.
Treat the months ahead as time to learn, not time to buy. If you have wanted a reason to get deliberate about which agents and which people can pull context from across your platform, this is it, and it is sound practice whether or not a bill ever arrives.
The teams who spend this window learning their own usage will not be the ones caught off guard.
From the Champion side of the table
This is a real change, and a little wariness is fair. Pricing that moves with usage feels less predictable than a seat you can count, and that feeling deserves an honest answer rather than a brush-off.
The honest answer is that the tools to see it, cap it, and forecast it sit in the same admin console as the meter, and Atlassian is running a customer webinar to walk through them. Start there, not with a worst case you imagined at the top of the page.
The meter did not change what makes AI worth using. It put a number on the context underneath it.
Understand the context, and the rest gets calmer.
📚 Further Reading
This piece builds on an earlier piece about MCP and Rovo, where the point was that the interface matters less than what moves through it. Usage-based pricing is that idea with a price tag on it.
- Atlassian, Usage-based pricing: the official source for the meters, allowances, rates, dates, and the customer webinar.
- The Unwritten Layer: the premise this piece takes as given, that AI is only as good as what sits underneath it.
- MCP and Rovo: the earlier read this one builds on, where the interface mattered less than what moves through it.
- AI That Works: the full series, if you are new here.
About the Author
Dave Rosenlund is an Atlassian Community Champion who writes about what it takes to make AI work in real enterprise environments. In his day job, he works for Platinum Atlassian Solution Partner, Trundl.
AI tools helped research, develop, draft, and refine this piece. The ideas, structure, judgment, and responsibility for what it says belong to the author. AI helped get past the blank page and supported the work that followed. The thinking is human.
Was this helpful?
Thanks!
Dave Rosenlund _Trundl_
5 comments