The great AI operating model experiment: What's changing in Tech?
The tech industry is running one giant AI experiment.
AI is changing the economics of building software, forcing companies to rethink how they structure product teams, where they invest and how they take products to market. I'm seeing smaller engineering teams, faster product discovery, fewer management layers and some expensive experiments in organisational design. But one thing hasn't changed: building something faster doesn't mean anyone wants it.
A lot of what I'm sharing comes from conversations with Executives, Product leaders, Engineering leaders, and others working through these changes firsthand. Some of the examples are commercially sensitive, so I won't name the organisations involved, but they're giving me a fascinating view into how differently companies are responding to AI.
1) Why the economics of software are changing
There has been a noticeable shift in the Australian tech industry lately.
Redundancies have become the new norm. Some companies are making cuts to survive because they never found a profitable business model during the ‘growth at all costs’ COVID boom. At the same time, investor appetite is shifting away from traditional SaaS towards AI-first and AI-enabled companies.
Others are restructuring because AI is reshaping how work is getting done, and lower-level tasks can increasingly be handled by AI with a more senior human in the loop. The question they’re asking is: how can we reshape our teams to take advantage of what this new technology enables?
We’re seeing the initial ‘all-in, everything-must-be-done-with-AI’ mindset shift towards more responsible usage, with some organisations now asking individuals to monitor token allocations because costs have blown out unexpectedly.
As one executive recently told me:
“We used to monitor whether people were using AI for 80% of their activities. Now we want that closer to 20% of their tasks. One thing is for sure: there's a noticeable difference in performance between those who have and haven't embraced AI. Finding the right balance between cost and the tasks that genuinely benefit from AI has been a challenge."
The shift is showing up in where investment is flowing too. Across the first half of 2026, AI has become an increasingly dominant theme in Australian venture funding. AI-first and AI-enabled companies accounted for 64% of venture deals in Q1, rising to 71% in Q2, while AI featured in 81% of Seed deals. Australian startups raised around $3.5 billion across the first half of the year, making it the second-strongest first half recorded by Cut Through since 2022.
The direction of travel is pretty clear: investors are increasingly backing businesses built around, or materially enabled by, AI, while becoming more cautious about traditional SaaS companies bolting on AI features to existing products.
When the cost of building software changes, everything around it has to change too.
How we structure teams. Where we invest. What we choose to build. And how we get those products into customers' hands.
2) The great operating model experiment
How are companies experimenting with their operating model and is it working? Well, the results have been varied.
I'm seeing engineering teams shrink, with much smaller teams using AI tools to build features in days that used to take months. Some are rebuilding their existing products to take better advantage of AI-enabled engineering.
One CTO I spoke with recently said a team of three engineers rebuilt their entire product in four months, the same product that had previously taken eight engineers four years to get to the same point.
Some have removed Product Designers altogether from the cross-functional teams, with Product Managers using tools like Figma Make or Lovable for rapid prototyping.
I’ve heard about companies that went all in on Product Managers at the expense of Engineering headcount, only to see disastrous results in production.
Conversely, Engineers are burning out because they're put directly in contact with stakeholders who have opened the feature floodgates, without Product Managers there to validate and prioritise based on whether there's a commercially viable need in the first place.
3) Product discovery is getting faster, but some are skipping it altogether
I’ve seen product discovery speed up to days vs weeks. A skilled discovery team can validate whether an idea is viable, desirable, usable and feasible faster than ever.
But I’ve also seen companies completely skip this step altogether, in a frenzied desperation to build everything they can as fast as they can to remain relevant.
As one founder explained their new approach:
“If a customer asks for a feature, we now build it straight away. We even have customer service team members making changes to the product in production.”
If AI can compress months of engineering into weeks, it can also compress weeks of discovery into days.
The coveted speed improvement that so many Agile transformations promised? It's finally here. But faster execution isn't an excuse to stop validating. It's an opportunity to validate faster.
4) The changing shape of the workforce
The impact isn't limited to how products are built. It's changing the structure of the organisations building them, and the career opportunities available to the people working inside them.
One of the more interesting changes I'm hearing about is companies reconsidering their offshore operating models. In some cases, work that would previously have been outsourced is now being handled internally using AI, reducing both the cost and the administrative overhead of managing offshore teams.
As companies downsize, they're also stripping out management layers. Founders are taking a more active role now that their start-ups or scale-ups have considerably shrunk, in some cases by half or more, and they're in survival mode.
The the same time, there are more senior product leaders in the market, but I’m also seeing a trend where companies take bets on promoting internal candidates into people leadership positions. Senior Product Managers now have more opportunities to step up, but they're also competing with peers who may get the role instead, leaving them reporting to someone who is learning how to lead on the job.
It's an interesting shift, and one that raises questions about how we're preparing experienced Product Managers to make the leap into people leadership. If we're promoting people into leadership earlier, how are we helping them learn to lead?
There are various reasons for this. Internal promotions bring some stability to a team already shaken by layoffs. It’s cheaper to hire internally than externally. And salary expectations can look very different when someone is stepping up into their first leadership role versus bringing years of leadership experience from the outside.
Removing management layers may reduce headcount, but it doesn't remove the work of leadership. It’s tempting to cut out middle management to give founders and executives more direct access to cross-functional teams, but people still need managers who coach, guide and support them. In one example, a founder increased his direct reports to 15+ people and replaced weekly 1:1s with ‘ad hoc’ communication. I was told team satisfaction and trust levels tanked.
I’m also seeing tech workers burn out and grow increasingly disenfranchised with the industry, not only because of the increased pressure on those who remain, but because of the lack of humanity and respect they’ve seen in how people are treated during layoffs.
5) Building software is getting easier. Getting customers is getting harder.
While companies are figuring out how to reorganise themselves, the market they're selling into is changing just as quickly.
Engineers are expected to pump out new features at the speed of light, sometimes without anyone stopping to consider whether there’s a genuine market for what’s being released. The result can be customers overwhelmed with features they don’t need or want, followed by price rises to pay for them. Unsurprisingly, some are taking to social media to vent their frustration.
With the AI shake-up, how tech companies operate is in uncharted territory, and many are racing to figure out what this means. Customers suddenly have more options too. In some cases, they can build solutions themselves rather than pay for another piece of software; meanwhile, vibe-coded apps are flooding the market. Everyone seems to have a vibe app these days, or at least an idea for one they’re wondering if they can monetise.
Which means the challenge is shifting. Building the product is getting easier. Standing out from the crowd, positioning it well and actually getting it into customers’ hands is becoming much harder.
6) AI makes execution cheaper, but judgement becomes more valuable
What does this all mean?
That the market has changed significantly and the old way of operating is fast being superseded.
Knowing how to work with AI is fast becoming table stakes. But you’re still running a business at the end of the day, and throwing AI-shaped spaghetti at a wall and hoping it sticks without a solid plan for whether there is a market and how you reach it is just going to result in a saucy wall.
I continue to see the argument that AI makes execution cheaper, so judgement becomes more valuable. But what does that rather vague statement actually mean?
Judgement is knowing where AI genuinely creates leverage and where it doesn’t. What should be automated and what still needs a human. Which problems are worth solving. Whether there’s enough evidence to place the bet. What not to build. And when to stop.
Just because you can, doesn’t mean you should.
Can remove designers ≠ should remove designers.
Can slash engineering ≠ should slash engineering.
Can generate 50 features ≠ should build 50 features.
Can vibe-code an app ≠ anyone wants the app.
Can automate offshore work ≠ every process should be automated.
Can use AI for discovery ≠ blindly trust its conclusions.
It’s tempting to be seduced by AI because it’s shiny and impressive, but unless your product is solving a real problem that is painful enough for people to pay for your solution, you’ve just got a rather expensive development hobby dressed up in a business suit.
7) Rethinking the traditional cross-functional team
And if better judgement, faster validation and stronger go-to-market capability are becoming more important, perhaps it's time to rethink how we structure the teams responsible for delivering them.
Your current product may no longer be relevant in this new landscape, and if you’ve put all your eggs in one basket, you’re putting your company at risk.
Your product portfolio should include more considered bets, and you’re better placed to make those bets if you’ve staffed for them properly.
It’s part of the thinking behind Bexlane’s Next Big Bet Accelerator. Rather than spending weeks building an idea before you know whether it has legs, a small cross-functional team from your business is guided through an AI-enabled process over 8–10 days to test the idea. Gather evidence and make a go/no-go decision on whether to invest further, and how you might take it to market. The benefit? An idea that has been floating around the business for months becomes something tangible you can put in front of customers, test and make a more informed investment decision about.
I think the traditional cross-functional software team needs to change. We may no longer need huge teams of engineers to build certain types of software, but we do need much stronger go-to-market capability, something that has traditionally been overlooked when it took so darn long just to build the product.
What might an AI-enabled product team look like?
Perhaps the next version isn’t a PM, Designer and eight Engineers.
Perhaps it’s a highly capable Product person, Designer and Engineer, all heavily AI-enabled, with Marketing and Sales involved from the start.
All sponsored by an Executive Leader and supported by Coaches who can upskill individuals as the work changes.
Getting to product release is no longer enough for product teams. When building becomes faster and cheaper, knowing how you’re going to take that product to market, get it into customers’ hands and grow it suddenly becomes liquid gold.
ABOUT THE AUTHOR
Bec Badcock
Founder & Fractional Chief Product Officer, Bexlane Advisory
Bec is a product executive and strategic adviser with over 15 years of experience building digital products, scaling product organisations and leading teams through growth and transformation.
As Chief Product Officer of a global SaaS provider, she led Product, Design, Marketing and Operations, including the organisation's transition from AI-curious to AI-native.
Her career spans start-ups, scale-ups and global enterprises. From launching Australia's first online supermarket marketplace at Woolworths to transforming product operations at THE ICONIC, she has delivered digital products and growth initiatives for brands including Qantas, Tesco UK and Subway US.
Today, through Bexlane Advisory, she works with founders and executives to shape product strategy, make better investment decisions and build high-performing product organisations, while developing the next generation of product leaders.
Published: September 2026
