SaaScceleration with AI: Notes from a Front Row Seat

From dashboards to daily decisions
One founder described how they effectively “fired” their dashboards.
Instead of logging into half a dozen tools each morning, their sales and marketing leadership team now gets a daily briefing compiled by AI agents plugged into all their customers calls and systems of record. The output is not more graphs; it’s a prioritised action list: which leads are heating up, which customers might churn, which renewals need attention.
Under the hood, they’ve simplified their stack dramatically, moving towards lean systems of record exposed via API, with AI doing the heavy lifting on top. The more interesting change is cultural: less time assembling information, more time making decisions.

Non technical builders are stepping forward
Another theme that kept coming up was the rise of the non technical “builder”.
One company has started a group for non engineers, armed with AI coding tools and a safe environment to ship internal automations. In just six weeks, they’ve automated many tasks, including a new internal service hub that previously would have sat at the bottom of the engineering backlog.

What struck me wasn’t the cost saving; it was the mindset shift – people who would never have called themselves “technical” a year ago are now building real tools that the rest of the organisation relies on.

3–10x speed, if you can handle it
On the engineering side, the numbers are getting hard to ignore. Founders talked about 3–10x speed ups in development cycles: user stories drafted in minutes, APIs rewritten by AI and reviewed by humans, legacy codebases refactored in days instead of months.

But nobody in the (virtual) room was under any illusions that this comes for free.
Several people warned about the “80% trap”: code that mostly works but hides deep, subtle issues. The response has been to double down on guardrails – automated scans on every pull request, strict security policies, and human QA that remains the final line of defence.

People who understand both system design and product management seem to be getting the best results with AI. That mix of skills could be a new competitive advantage.

The net effect is interesting: writing code is no longer the bottleneck. The bottleneck has moved to testing, QA, security, and organisational learning.

When internal tools become the product
One of my favourite stories came from a logistics software solution.
They started by building AI tools to help their own team troubleshoot customer tickets. Over time, that internal capability evolved into a customer facing AI assistant directly inside the product. Today, a customer can type “reschedule this delivery next week” and the AI assistant orchestrates all the underlying steps – something that used to require a lot of manual clicking.
That’s the kind of change AI makes possible when it moves from “tool for the team” to “feature in the product”.

Beyond faster code
The most pointed challenge to today’s AI narrative came from a founder working on complex capital projects.

The argument was simple: writing code was never the real problem.
The hard part is understanding which problems matter, exploring the space of possible solutions, and getting changes into the hands of customers safely. If all we do is speed up the middle of that cycle – the coding bit – we can actually make things worse.

Their team is experimenting with “code aware requirements”: product owners write requirements that are directly informed by the current state of the codebase, with AI helping surface constraints and trade offs. They’re also feeding behavioural analytics into problem discovery, so that AI helps them decide what to build, not just how to build it faster.

It’s a subtle point, but an important one: the real opportunity isn’t AI as a pair programmer; it’s AI as an engine to redesign the entire product lifecycle.

Why we’re optimistic
After the session, I jotted down a few patterns I’m seeing again and again.
• The technology stack is quietly unbundling: strong systems of record at the core; AI providing the intelligence, automation, and orchestration around them.
• Non technical talent is being genuinely empowered. With the right context layer and deployment platform, people who never considered themselves “technical” are now building useful tools.
• The bottlenecks have shifted. Speed in coding is abundant, with in some cases around 95% of code written by AI; quality, security, and learning are scarce. The winning teams are reallocating energy accordingly.

And perhaps my favourite observation: today is the worst these AI models will ever be.
Models are improving week by week. The constraint is no longer raw capability; it’s our ability to adapt our organisations, our processes, and in many cases our business models to take advantage of what’s already on the table.

From where we sit as investors, this doesn’t look like a “SaaSpocalypse”. It looks like “SaaSccleration”, a rare moment where ambitious and innovative software businesses can compound advantages faster than before – if they’re willing to rethink how they work.

At ETF Partners, we back entrepreneurs who use technology – including AI – to build businesses that can make a positive environmental impact. Conversations like this give us confidence that innovative software has a central role to play in that transition.