By Krista Hulm, APSJobs and Digital Services, APS Academy
Finding the right learning opportunity in the APS can feel overwhelming. Learning options are spread across multiple systems and platforms. Time is limited. And knowing what's genuinely relevant to your role, your level and your career goals takes effort.
The APS Academy is exploring whether AI can help solve this problem through a proof of concept called APS CoLearn.
What is APS CoLearn?
APS CoLearn is a GenAI-powered learning assistant designed to help APS staff discover trusted learning opportunities, build personalised learning plans and get guidance tailored to their career goals in a conversational interface.
Rather than replacing existing learning platforms, CoLearn acts as a guide across the learning ecosystem, surfacing APS Academy content alongside other vetted options and helping staff make sense of what's available to them.
The project is a small, time-limited proof of concept. It's focused on testing whether this approach genuinely helps people find learning and increase confidence in their decisions.
What we're hearing from users
Early user research has been encouraging. In testing with APS staff from several agencies, participants shared they would choose CoLearn over the current search experience for finding learning.
One participant shared:
‘It's a lot quicker than it would be if I was searching manually.’
The tool's strongest value isn't just surfacing courses. It's structuring a learning pathway. Participants responded to features like learning plans, sequenced recommendations and suggestions they could take to a conversation with their manager.
Staff want recommendations that reflect the APS context, not generic content designed for the private sector. They want personalisation that's low-effort where the tool learns about them through conversation, not lengthy forms.
As one participant put it:
‘I'm looking at the citations and... they're all .govs... We haven't gotten a random stack of information.’
What we're learning along the way
Our team has been building to learn. Some key takeaways so far:
- Better data produces better AI. The quality of what the tool recommends depends directly on the quality of the information it draws from.
- Trust is built in the first interaction. If early recommendations feel generic or link to the wrong thing, users disengage quickly. Context, transparency and relevance matter from the start.
- AI is probabilistic. Unlike traditional software, results are not predictable. This makes testing and iteration more important than ever.
- Reusable infrastructure helps. Access to shared AI platforms like GovAI helped the team experiment faster without building everything from scratch.
- Learning by doing builds capability. The project is growing the team's AI engineering skills. This is a benefit that extends beyond this single use case.
What's next
The team is continuing to refine and test the prototype through to October 2026. A report on outcomes and feasibility will inform whether CoLearn moves toward production.
Get involved
If you're interested in learning more about APS CoLearn, or would like an invitation to our end-of-project showcase in October, get in touch.