A convincing coding demonstration can be short. Running that capability across a real company introduces a much longer list of questions: which repository matters, whose rules apply, and who can authorise the next action?
On September 10, Atlassian announced capabilities intended to support governed agent workflows across the software development lifecycle. Its announcement described Code Context, built on the Teamwork Graph, alongside Agent Context Controls. The company said the former provides intelligence across code repositories and the latter allows platform teams to control where agents operate and what information they can access.
GlobalFounderTimes analysis: the distinction between capability and permission is central. An agent might be able to complete a task while still being the wrong actor to perform it in a particular environment. A useful enterprise product has to address both questions. The impressive output is only one part of the purchase decision.
Context presents another challenge. A request to fix a bug may depend on an earlier architectural choice, a customer commitment or a service that lives in another repository. Treating each instruction as an isolated prompt can miss the reason a system was built that way. Making context available is valuable only if the information is current and the agent has the appropriate access.
For founders selling AI tools, this suggests a more concrete demonstration. Show a task the agent should complete, a boundary it must respect and an escalation it should send to a person. Buyers can then assess behaviour under constraints rather than judging a perfect-path example alone.
There is an important limit to the announcement: a vendor’s stated control features do not independently prove reliable outcomes in every deployment. Teams still need to examine their own permissions, review procedures and failure handling. Atlassian’s September release is interesting because it frames those operational concerns as part of the product. As agents become more capable, the ability to explain why they acted may matter as much as the action itself.