AI capabilities in product development have leaped forward in 2026 in ways that we are only just comprehending and as an advisor to tech orgs in Australia, I've been wondering - how much of the hype is actually true? And what has enabled these huge productivity gains? Rather than reading the sensationalist articles, I reached out to a few of my friends that work in big tech, senior engineers and product managers on the front lines of AI adoption, to try to make sense of what's working and importantly, why. I've distilled my learnings from these conversations into a 6 part series, each comprised of an insight, a practical takeaway and considerations for Australian organisations based on what I've seen historically play out in the industry.
The productivity range is wider than most people expect. One engineer estimated he's 10-20% faster on his day-to-day work. Meanwhile, a product manager described multiple teams completing their entire 6-month roadmap in a two-week sprint - and this wasn't a one-off; it happened across at least four or five teams and went viral internally. What explains the gap?
For day-to-day work, the clearest gains are in the mechanical stuff - boilerplate, refactoring, style fixes, linter errors. One engineer described handing off 150 style fixes that would have taken 30+ minutes; the AI handled them in just a few minutes. But the teams that compressed 6 months into 2 weeks weren't grinding through boilerplate. They locked themselves in a room for two weeks with a simple question: how much of our 6-month roadmap can we get done with AI? At least four or five teams came out the other side having completed all of it. As one PM observing this put it: "This is crazy."
This example is also completely unsustainable - these teams were crunching 12+ hour days, just to see if they could pull off such a monumental feat. They were also physically in the same room together, much like a hackathon. But the harder work - architectural decisions, system design, cross-team alignment, figuring out what to build and why - remains largely unaffected. As one engineer put it: "You still have a lot of time in the non-coding pieces - design and planning and syncing with other people." This seems like 2 weeks of work, but that misses the months of conversations and coordination that went into landing on this particular 6 month roadmap.
Set expectations that match reality. The headline-grabbing stories are... somewhat real, but they happen when teams with strong foundations go heads-down on execution-heavy work with clear direction and alignment. For most teams, on most days, the gain is meaningful but not transformational. And not all engineering problems are ripe for such big gains - one engineer working on a very complex system stated that they still hand write 50% of their code and quoted gains of approx 20%.
Frame AI as a tool that compresses execution - and then redirect the conversation to what your organisation will do with the time it frees up.
If you don't have a well-informed product strategy or discovery practices, this becomes a bigger problem than it sounds. The freed-up capacity needs somewhere useful to go. In organisations with clear strategic direction, teams can redirect it toward discovery, experimentation, or harder problems. Without that direction, it just gets filled with more output - more features, more tickets, more "busy work" - and nobody asks whether any of it matters. The PM I spoke with reflected on the teams that shipped their roadmaps so quickly: "More importantly, are they working on the right thing? I don't know."
Stay tuned for Part 2: Quality is the new bottleneck
