The honest bit: AI is only as good as the estate under it
Every Atlassian AI feature reads your estate. Rovo's answers draw on your Confluence — if the knowledge base is a landfill, the answers inherit the smell. Agents act through your projects and workflows — if categorisation is chaotic and permissions are folklore, you've automated the chaos. This is why "just switch it on" disappoints, and why the organisations getting real value treated AI adoption as the reward for an estate in good order.
What it costs — the honest shape
Atlassian's AI pricing has been moving: capability bundled into higher plan tiers, and usage-based elements arriving alongside. We won't print numbers that stale by the next quarter — our analysis of the usage-based pricing shift covers the model — but the question that matters is never the list price: it's what it costs you, on your plans, at your scale, against your actual usage. That's a licence-position analysis, and it's the same audit muscle we run on the app estate.
Agents, and connecting them: MCP
The frontier isn't chat — it's agentic AI: assistants that act in your service desk, and external AI agents reaching into Jira and Confluence through MCP (Model Context Protocol), which Atlassian now supports. Done well, that's tickets triaged, knowledge answered and work moved without a human relay. Done carelessly, it's an ungoverned actor with admin-shaped access. We build agentic integrations ourselves — the governance questions (what can it see, what can it touch, what gets logged) are exactly the ones we help you answer before anything gets switched on.
How we help
Find out where you stand: the JSM Health Check already asks whether your AI capability is doing anything useful — and whether the estate underneath it (knowledge, categorisation, permissions) would let it. Adopt at the platform's pace: on PlatformCare, each release's AI capability is discussed against the business problems you actually have and implemented where it earns its place. Or take it as a project: evaluate, prepare the estate, configure, govern, train — scoped and priced from a short assessment.
One promise
We'll tell you when the answer is "not yet." An AI feature switched on over a weak knowledge base doesn't just underperform — it teaches your teams to ignore it, which is harder to undo than never enabling it. Sequencing is most of the value, and sequencing is judgement work.
Is Atlassian's AI doing anything useful for you — and if not, what would it take?
Book a 30-min consult
No deck. No sales pitch. Just a useful chat.
Tell us roughly what you’re facing. We’ll come back within one business day with thoughts, references, and a sense of whether we’re a good fit.
- ✓ Free, no obligation
- ✓ Direct line to a senior consultant
- ✓ Honest assessment — we’ll say no if it’s not a fit