CBA using AI as part of its ‘big room planning’

CBA is using AI as part of its ‘big room planning’, a quarterly Agile activity that brings 16,000 people from across the bank together to coordinate software delivery activities and goals.

The bank first introduced Atlassian Intelligence – AI designed to interact with a company’s Atlassian Cloud environment – to support users in the big room planning process in December last year.

Engineering platform general manager Helen Lau told Atlassian’s Team ‘25 conference that the bank had laid the foundations for this by migrating its Jira and Confluence environments from on-premises versions to the cloud over the past few years.

Both tools are already part of the big room planning process; Jira, in particular, because it maps and tracks dependencies – inputs or contributions required from other teams – needed as part of a project or program of work.

A significant part of the big room planning process is to put “everything worth a discussion on the table”, Lau said.

Issues that are contained to specific teams are dealt with by the teams themselves.

“For anything internal, your team has full autonomy to work out what you need to do, but when you depend on someone to lift a finger or you have a lot of demand coming in, you need to have that conversation: Can I deliver this for you? Can you deliver this for me?” Lau said.

Big room planning is used to get those issues and discussions out in the open, determine what work is prioritised and funded over the next quarter, and conversely, what can be deprioritised. 

Lau said the bank had so far gone through “two cycles of big room planning”, with 16,000 people from its Australian and Indian operations coming together.

AI is being used as a support in this process primarily to summarise lengthy materials and get to the crux of a conversation much faster.

Lau said that Jira dependency tickets were often brief but pointed to detailed documentation in Confluence.

AI is used to summarise that documentation.

“We use that a lot to help us get to the key content and message,” Lau said.

“We’re using AI to help us quickly get to the content to have the right conversation as early as we can, rather than reading a document, [sending an] email, [having a] meeting, [sending another] email, [having another] meeting.”

Atlassian Intelligence is being used more broadly than just the quarterly planning process.

Lau said CBA is now saving “around 2500 hours a month on summarising key delivery documents”.

It’s also helping squads at the bank, which comprise between 10 and 20 people, to “break down epics into user stories”.

Within Agile, an epic “is a body of work that can be broken down into specific tasks called user stories”, according to Atlassian documentation.

“Each squad saves about 14 hours a month now in terms of breaking down an epic into the user stories,” Lau said.

“That’s across 1100 squads in the group, so that’s a big number there.”

She added that “internal customer satisfaction” in being able to use AI on information in the Atlassian tool ecosystem is also “the biggest cherry on top for us”.

“We get a lot of [internal] customers coming back to say they love it,” she said. 

“When you work in IT, we don’t get much ‘we love it’ feedback. We get a lot of P1, P2 [priority incidents]: ‘Can you fix this?’

“So, customer satisfaction was a big bonus for us.”

Big room planning, as part of the bank’s Agile and DevSecOps embrace, is contributing to CBA getting more new products and features into production, which can benefit its customers.

“When I joined the bank three years ago, we were making about 2000-to-3000 changes a month,” Lau said.

“Now, we just ticked over 9000 changes a month in production, meaning 9000 features a month went into our customers’ hands. 

“It could be a prompt, a chatbot, or another feature. So that’s the efficiency gain and velocity we’re talking about in terms of outcome.”