Chemist Warehouse's AI tool for HR becoming a "standard pattern"

Chemist Warehouse appears set to expand its use of an AI-based shared inbox automation tool, with an initial deployment in its HR function now acting as a model for other internal functions to adopt the technology.

Speaking on a Microsoft-run webinar, Insurgence managing director Matteo Castiello said that what was built for the HR function “is actually a very common challenge …for many other teams across Chemist Warehouse.”

Insurgence is the Microsoft partner that worked alongside Chemist Warehouse on the AI tool - called the AI Human Resources Advisory or AIHRA - which drafts email responses to common questions from the retailer’s 22,000 staff that are sent to a central inbox.

Castiello said the HR function had established itself as a kind of “incubator” for the technology internally.

“We’ve started to really get more exposure across the business, with AIHRA now becoming a standard pattern,” Castiello said.

“It's becoming a standard shared inbox automation tool that any team can pick up and start to get some efficiencies in all [internal] domains.

"The challenges [in these other domains] are the same, which means that the functionality is the same and it meant that a lot of what we're doing now in HR is acting as an incubator for all of these other teams that want [a] more general purpose shared inbox automation tool.”

Chemist Warehouse deployed the AI agent into its people and culture function earlier this year. In part, it was intended to stop its HR staff burning out and leaving due to highly repetitive workloads.

Head of people and culture advisory Melissa Bear told the webinar that the impetus for the project was “a turnover issue in our team”.

“That's a tough thing obviously to talk about, but the main reason for the exits from our team was the repetitive nature of the work and the high volume,” Bear said.

“We were finding that the team was starting to feel a bit burnt out and not enjoying the work quite as much as we would've liked. 

“And of course, keeping key staff and their capability is really important to the work that we do in the HR field.”

Chemist Warehouse's people and culture shared inbox is the first point of contact for HR-related support nationally.

Bear described the shared inbox “as kind of the highway for [staff] to quickly get to us with their needs and inquiries.”

Aware of the repetitive nature of “low risk” queries, Bear said she initially drafted some boilerplate responses to the most common questions. HR advisors could manually copy it, pasting in response and tweaking as necessary.

“That only kept us going for a little while. Then it was about [finding] a much smarter solution to that,” she said.

“That's where AI has come into the picture for us – to actually start to create a bank of knowledge that informs the AI solution, that then starts to draft email responses for us in our national inbox to selected topics that we've built that content for.”

The AI agent scans the shared inbox every 30 seconds, picking up inquiries it can respond to and then drafting a response that is left for a HR advisor to review.

“The draft gets formulated, it sits in our national inbox, and the relevant advisor or senior advisor will pick that up, review the draft [and] make sure it's fit for purpose because our AI is still learning,” Bear said.

“We get to do that refinement if it's needed, but most of the time for the 10 topics that we've built in terms of knowledge library, it's pretty much click and send.”

Building the knowledge bank that the AI agent could reference involved some considerable heavy lifting.

“For Sigma Healthcare-Chemist Warehouse, this company's growing exponentially and so quickly that some of our ways of working and knowhow just simply weren't documented,” Bear said.

“The confronting part [of the project was] how much knowledge is in our heads [and] not documented anywhere. 

“So, a challenge was to really get that down because that is the golden content that informs the draft response [by the AI agent].”

Castiello said the “initial version of AIHRA used a combination of Power Platform and Copilot Studio, combined with Azure AI Foundry for more of the complex work.”

Both Castiello and Bear said that significant effort was also invested into ensuring the AI stuck to its lane.

“We probably spent more time trying to teach AIHRA what not to respond to than what to respond to,” Castiello said.

Bear said that people and culture staff were so far pleased with the support provided by the tool.

"They [now] have more time for more important work, which is coaching our leaders, supporting employees in times of difficulty, going through workplace investigations, mitigating risk, ensuring we're compliant, and just continuing to build and maintain our brand and our credibility as a HR service,” she said.