“AI is eating software”? No. Software is eating AI

The rise of Agentic AI and plugins from parties such as Anthropic and OpenAI is putting pressure on the traditional SaaS model. Functionalities that for years formed the foundation of software products suddenly seem generically available. Yet, this doesn't mark the end of software, but the beginning of a new phase.

Lately, you hear it more and more often: “AI is eating software.” Within the world of Software as a Service (SaaS), this statement is causing unease. Companies that have successfully developed and delivered applications via subscription models for years are seeing AI players such as Anthropic and OpenAI rapidly launch plugins and agentic applications that seemingly take over functionalities around which complete SaaS solutions were built. What was yesterday a distinctive product, today appears available via an AI assistant performing the same task without the user needing a separate tool.

It is understandable that this raises questions about the legitimacy of traditional software. If an AI assistant can create reports, perform analyses, structure data, and even automate processes, where then is the value of specialised software? As a result, many SaaS companies see their business model under pressure. Functionalities that were premium for years threaten to be commoditised by generic AI solutions.

Yet, I believe that the conclusion that AI “eats” software is too simplistic. In fact, I believe we are moving towards the opposite. Not AI replacing software, but software absorbing AI. In other words: software is eating AI.

Software is not losing its right to exist. It is gaining a new dimension. Where software used to be about functionality, workflows, and data processing, it is now increasingly about intelligence, adaptability, and context. AI is not a replacement for software, but an enhancer of it. The software that will be successful in the coming years is not the software that tries to ignore or fight AI, but the software that deeply integrates AI into its core.

The difference lies in how you apply AI. If AI is just a standalone feature, for example, a chatbot button in an existing application, fundamentally little changes. But when AI is used as the engine behind the product, the nature of the software itself changes. Then you shift from static systems to dynamic systems that continuously learn, monitor, and optimise.

A concrete example of this is Proxies, a company we recently started. At Proxies, we build highly detailed company databases for Europe. Traditional databases are essentially static. They contain snapshots of information: revenue figures, number of FTEs, sector classifications, and other company characteristics. Updates are often periodic, depending on fixed sources or manual input. The information is valuable, but always a snapshot.

With the use of Agentic AI, this fundamentally changes. Instead of a database that merely stores what is known, we create a system that actively gathers and understands information. Our AI Agents act like digital researchers. Just like a human, they specifically search for relevant information about companies on the web and in public data sources. They analyse context, recognise signs of change, and continuously monitor developments.

Consider, for example, changes in the number of employees, revenue growth or decline, new branches, strategic announcements, or management changes. Where this was previously only possible with large analyst teams or limited sampling, it can now be done scalably and continuously thanks to Agentic AI. The system doesn't wait for a quarterly update; it actively monitors what is happening.

This was simply not possible before. Not because there wasn't a need, but because the technology was lacking to gather contextual and intelligent information at scale. Agentic AI makes it possible to build software that not only manages data but actively generates knowledge.

And here lies the core of my argument. Generic AI solutions are powerful, but they are broad and general. They do not know the nuance of a specific market, sector, or use case at the level that specialised software can. The real value arises when you combine AI with deep domain understanding, structured data, and a clear product vision. Then AI does not become a replacement for software, but an infrastructure layer within software.

SaaS companies are therefore not facing an existential crisis, but a strategic choice. Do you stick to a model where functionality is central and AI is seen as a threat? Or do you redefine your product by integrating AI as a core component and elevate your solution to a higher level?

The possibilities created by plugins, agentic systems, and new AI tools are immense. They make it possible to build software that is proactive instead of reactive, intelligent instead of static, and adaptive instead of rigid. This demands a redesign of products, a reconsideration of business models, and above all, courage.

Therefore, I want to challenge companies not to react defensively to the rise of AI, but to think offensively. Don't ask “how do we protect ourselves against AI?”, but “how can our software absorb AI and thereby become better than ever?”

Because the level of software is not going down because of AI. It's going up. Much higher than we previously thought possible.

Want to learn and know more about Proxies? Check proxies.ai.nl

Want to get started with AI yourself? Check out our e-learning AI Agents or AI workshops for teams.

• Best AI Tools