Quicksilver Labs

Why This Matters

We are in a market where valuation and capital availability is all about AI. Investors have zero interest in non-AI software companies and are deeply skeptical about legacy software vendors even when they report to be using AI.
Enterprise build-buy and AI-Native (startup) options will put serious pressure on the traditional SaaS business model. Buyers are willing to consider replacement of legacy software to get the advantage of being valued for AI themselves.
Companies that fail to make true AI-Transformation NOW risk not having resources to do so later and therefore risk their very survival.

The Valuation Imperative for Legacy Software

Enterprise Vertical SaaS valuation multiples are down by 30%, or more, over the last year (AKA "SaaS-pocalypse"). The valuation drop in private companies that are non-AI native ("legacy software" for lack of a better term) is even larger than for public companies. These companies urgently need to prove to the market that they are not vulnerable to AI-startups or internal, AI enabled "build vs buy" decisions. Doing this is hard:

Transition Requires Deep(er) Understanding of the Customer Workflow

Simply automating some portion of white-collar work away with LLM calls is straightforward and creates no real moat. However, we believe that transformation of workflow, new approaches enabled by AI, is likely to deliver the best payoff to the buyer and a bigger moat for the vendor. Startups are rethinking the entire workflow (ex: VOZE, a CRM with mobile & multi-modal first). This requires a level of customer understanding that many legacy software companies don't really have. One reason that every startup is talking about forward deployed engineers is that the FDE gets under the tent to really understand, and stimulate thinking about, where major benefits of AI can accrue.

Buyers Expect That Products Are Built for Their Industry

Because of the expectation that development is fast and cheap, enterprise buyers are going to demand customizability; applications that were horizontal are at risk of being displaced by vertical, industry specific applications. After all, the buyer can vibe code something that looks pretty much like what they want. Why can't the software vendor deliver it? Not that we believe that most companies will choose to build and maintain their own core systems, but the possibility will create additional expectations.

New Business Models Are Required to Align with Costs and Value

AI inference may increase the cost of delivery so software vendors must charge more or rethink their business model. Startups (and perhaps more flexible incumbents) are testing new ways to charge for value, aligning with benefits. An important part of any new business model is what moat it creates since swapping out low-moat systems in the future will become easier for the buyer.

Buyers Will Demand the Ability to Keep Up with Change

Buyers see AI capabilities changing rapidly. It seems like a long time ago that we had the DeepSeek moment but that was only a year ago. Now we just had the OpenClaw moment, the Co-work moment and the Citrini ("sci-fi") moment. Buyers know that whatever they purchase today could be obsolete before fully deployed. We think this same concern will limit internal "build"; very few companies can afford to keep a top AI team, albeit small, continuously developing, testing and deploying secure and scalable software. Buyers are extraordinarily open to purchasing new technology now but need to trust that the vendor can keep up with technology changes and deploy new advances to the benefit of the enterprise in a secure and scalable manner.

Slow Movers Risk Never Catching Up

Large enterprise SaaS companies are probably not going to be replaced wholesale in the next few years but, even for them, change is going to happen more quickly than in the past because the buyers of software are themselves being pushed by the market to prove AI based transformation (and benefits).

For smaller, private growth stage or earlier SaaS companies, downward valuation pressure will start to reduce fund-raising ability, potentially choking off the funds needed to make the AI transition. Many of these are just growing into profitability, are not so deeply embedded in the workflows as the major systems of record, and need to move fast to avoid the trap of reduced valuation making it harder to transform in the future and harder to be acquired or go public.