How AI is Rewriting the VC Playbook for Software Investing

The rapid evolution of AI has turned the traditional venture playbook on its head, replacing “solid fundamentals” with a brutal new standard for survivability. 

In this Q&A, Simone Riva, Partner at Partech, discusses the shifting landscape of enterprise software.

Simone brings a wealth of seasoned perspective to the table; prior to joining Partech, he served as Head of Digital Investments at H14, a leading Italian Family Office, following a foundational career in Private Equity at Bain & Co.  Having spent over a decade backing companies across the US and Europe, he has a front-row seat to the current structural shift.

From the erosion of software moats to the build vs. buy dilemma facing CTOs in 2026, Simone explores why code velocity is no longer enough, diving into the cultural divide between high-adoption firms and those lagging behind.

For any founder or investor, this is a look at what it takes to remain relevant when the bar is being raised ten times over.

AI is moving fast and valuations are high. As a VC, how has that changed the bar for what you’re willing to back?

Five or ten years ago, you could back a SaaS company with solid fundamentals and feel

reasonably confident it would still be relevant at exit. That’s no longer true. 

The speed at which AI is moving means the threat landscape changes faster than any investment thesis can anticipate. So I need to be ten times more precise in how I assess survivability – not just “is this a good company today”; but “will this company still exist in three or five years?”.

That bar has been raised significantly. At the same time, entry valuations are high and Limited Partners are pushing hard for liquidity.

AI is making it cheaper and faster to build software internally. What does that mean for the traditional software moat?

It’s a real structural shift, and I think some investors are still underestimating it. We’re already seeing large enterprises replace SaaS products – tools at the level of Salesforce – with things built internally using AI. That’s not a marginal trend. 

For SMBs it’s less pronounced because they often don’t have the engineering capacity (and in some cases are still on Excel), but for mid-to-large enterprises, the calculation is changing. 

In my opinion, the companies most at risk are those in which the product sits on top of workflows rather than inside them. 

If a company’s core value is essentially automating a task that an engineering team can now replicate in a few days, that moat has largely gone.

The companies I want to back are the ones where switching is genuinely painful: deep workflow integration, proprietary data, or distribution advantages that can’t be rebuilt quickly.

Where is AI actually delivering in enterprise software today, versus where is the hype still running ahead of reality?

The three use cases I see working reliably in our portfolio right now: replacing junior developer tasks in specific contexts, API integrations that streamline operations, and customer service and support automation. Those are genuinely powerful. 

Where I think people are overestimating AI’s impact is around the speed and universality of adoption.

Companies founded in the last two or three years are much faster at shipping AI products than those founded in 2015 or 2016 – the DNA is just different. 

We actually saw this up close: Partech recently brought together around 40 senior CMOs and CTOs from across our portfolio and network, and one of the clearest signals from the room was that the gap between high-adoption and low-adoption companies has almost nothing to do with the tools available and everything to do with culture and intent.

There are also risks that don’t get enough airtime such as hallucinations creating liability issues, security vulnerabilities, vendor lock-in to a single model provider. 

Code velocity has never been higher, but the key challenge has shifted from writing code to reviewing it – and that’s a new kind of problem. The discipline of reviewing and auditing what AI builds is now a clear competitive differentiator.

For enterprise buyers trying to choose vendors right now, what’s your advice?

The build vs buy question has genuinely become more complicated. 

AI has lowered the barrier to building, which means vendors that were previously the obvious choice now need to prove their ongoing value more clearly. My honest advice is to prioritise depth of integration over breadth of features. 

A product that is deeply embedded in how your team actually works is harder to displace, both by a competitor and by something you build internally. And be honest about your internal capacity. 

Most organisations don’t have the engineering resources or the ongoing maintenance bandwidth to build and sustain something properly. 

The total cost of ownership of an internally-built solution is almost always higher than it first appears. The best vendors in 2026 are the ones who understand this, and can demonstrate clearly why staying with them costs less than the alternative – in time, risk and distraction.