Is AI eating software?

The phrase “AI is eating software” refers to how AI systems are rapidly transforming or even replacing traditional software development and operational processes. This concept mirrors Marc Andreessen’s famous quote from 2011, “software is eating the world,” in which he pointed out how software was becoming central to virtually every industry. Now, AI is beginning to take over many of the roles software has traditionally filled.

Software Eats the World: A Retrospective

When Marc Andreessen famously declared that “software is eating the world,” he was referring to the massive shifts in industries—retail, media, entertainment, telecom, and beyond—driven by software innovation. Companies like Amazon, Netflix, and Spotify transformed sectors by leveraging software to create platforms that scale globally without the physical constraints of traditional businesses.

This transformation was fueled by advancements in computing power, the rise of the internet, and the democratization of software development. Industries that were once anchored by physical goods or real-world services found themselves overtaken by digital solutions. Amazon, for instance, didn’t just replace bookstores with an online catalog—it built an entire infrastructure of software that allows it to sell nearly everything.

AI Eats Software: A New Paradigm

Today, AI is pushing this disruption even further, creating a scenario where AI is not just enhancing software but replacing it. Klarna’s approach of using AI to replace traditional SaaS solutions—from customer service to internal knowledge management—is an example of how AI is moving beyond software augmentation to outright dominance. Klarna’s AI-driven tools are performing tasks that once required dedicated software platforms like Salesforce and Workday, making the need for SaaS systems increasingly obsolete .

Just as Amazon transformed retail with software, AI is now transforming the very structure of how software itself is used, built, and integrated. AI systems like ChatGPT are not just automating code writing but providing dynamic, adaptive, and context-aware solutions that can manage tasks traditionally handled by separate software applications.

From SaaS to AI-First: Klarna’s Transformation

Klarna’s journey offers a glimpse into what an AI-first architecture can achieve. Instead of relying on external SaaS platforms like Salesforce and Workday, Klarna developed internal, AI-driven systems that manage critical operations, such as customer service, marketing, and legal processes.

This approach allows Klarna to integrate generative AI at the core of their operations. For instance, Klarna’s internal AI assistant, Kiki, answers more than 2,000 employee queries daily within seconds, significantly reducing the need for traditional knowledge management systems. Their AI-powered customer service has reduced resolution times from 11 minutes to under 2 minutes, further underscoring the transformative impact of AI on operational efficiency.

What is AI-First Architecture?

An AI-first architecture represents a fundamental shift in how software is designed and developed. Instead of adding AI capabilities to an existing framework, AI is deeply integrated into the system from the ground up, allowing for adaptive, learning-based functionalities that evolve with data.

Here’s what defines an AI-first architecture:

Core Integration: AI is embedded into the software’s core, making it an essential driver of functionality, insights, and user interaction.
Adaptive Design: AI models within this architecture are built to learn and adapt, rather than simply execute pre-programmed instructions. This allows the system to adjust to new data and contexts dynamically.
Comprehensive Components: A typical AI-first system incorporates a Cognitive API Layer, enabling easy integration of AI models, an Autonomous Agent Apps Layer that allows AI to interact with APIs and tools, and Guardrails to manage and optimize AI’s performance.
Data-Centric Approach: These architectures prioritize the use of data to continuously improve AI models, creating a feedback loop where AI learns from every interaction.

Klarna’s AI-First Components in Action

Klarna’s AI-powered systems extend across various departments:

• Communications: Klarna’s team uses ChatGPT to analyze media sentiment, allowing the company to respond strategically to public perception. This AI-driven solution is faster and more accurate than traditional media monitoring tools. Filippa Bolz, head of communications at Klarna, states: “Media monitoring companies have tried for at least fifteen years to automate ‘sentiment’ analysis but often you’d get better results from blindfolded monkeys throwing darts. The tool we’ve built using ChatGPT has blown our minds! It provides a really objective analysis of each article in a couple of seconds, which helps us ensure we communicate clearly, that it is understood and resonates with each one of our target audiences – rather than bundling them together.”

• Legal: Klarna’s legal department now drafts common contracts using ChatGPT Enterprise, reducing the drafting process from an hour to mere minutes. Instead of starting from scratch, legal teams tweak AI-generated drafts to meet their specific needs. Selma Bogren, senior managing legal counsel at Klarna, states: “The big law firms have had a really great business just from providing templates for common types of contract. But ChatGPT is even better than a template because you can create something quite bespoke. Instead of spending an hour starting a contract from scratch or working from a template, I can tweak a ChatGPT draft in about ten minutes. You still need to adapt it to make it work for your particular case but instead of an hour you can draft a contract in ten minutes.”

Challenges and the Road Ahead

While an AI-first architecture brings unprecedented efficiency, it also poses challenges. Developing custom AI systems requires substantial upfront investment and demands a shift in company culture to ensure that employees are equipped to harness the full power of AI. However, the long-term gains in productivity and cost savings make this approach attractive for enterprises looking to stay competitive.

The Future of AI in Enterprise Tech

Klarna’s strategic shift to an AI-first architecture is a clear indicator of where enterprise technology is headed. As more companies recognize the potential of AI to replace traditional software solutions, the AI-first model will likely become the new standard in industries ranging from fintech to healthcare. This evolution signals a future where AI doesn’t just assist with tasks—it is the software, driving businesses to operate smarter, faster, and more effectively than ever before.