The End of Human-Intensive Software Development: Navigating the AI World

Paik argues that SaaS, ARR, and traditional software metrics are remnants of a business model where high software creation costs acted as a protective moat for companies. With the rise of AI and large language models (LLMs) like GPT, this moat is quickly disappearing, and with it, the way we approach software development is fundamentally shifting.

At Bloomhaus Ventures, we see this change as not just inevitable but transformative for the entire industry. The shift away from human-intensive software development is already happening. Companies like Microsoft and ServiceNow are reporting that up to 50% of their code is now written by AI, while Amazon saved an estimated $700 million by using AI to refactor code. As this trend continues, the cost of creating software will drop dramatically, leading to a new era where software development is faster, cheaper, and more accessible than ever before.

‍

The Cambrian Explosion of Software

Historically, software development has been expensive because developers are expensive. The ability to translate human ideas into computer code requires years of training and expertise. Businesses have been paying for that expertise through licensing fees, subscription models, and per-seat pricing, creating a highly profitable industry with low distribution costs and high margins.

But with the advent of AI-driven tools like LLMs, the cost of software development is rapidly approaching zero. This change will likely spark a so-called "Cambrian explosion" of software development, similar to the explosion of user-generated content that followed the rise of the internet. Thousands of specialized software solutions will emerge, replacing the monolithic giants like Oracle and SAP. The protective moat of high software creation costs will vanish, making it difficult for any single company to dominate a market in the way traditional software behemoths have.

One clear example of this shift is Klarna, the Swedish fintech giant, which has been making headlines with its decision to abandon industry-standard software like Salesforce and Workday in favor of building its own AI-powered internal systems. After achieving success by automating two-thirds of its customer inquiries with AI, Klarna is doubling down on this strategy, betting that AI-enabled software is the future of internal tools.

Klarna’s move highlights a key trend: the overall cost of building internal software with AI is lower than purchasing off-the-shelf solutions.

By building customized AI-powered systems, they are effectively rearchitecting their entire software infrastructure, shifting from traditional enterprise software models to a new architecture that looks something like this: data lake → AI → bespoke software. In this model, AI isn't just a tool to help build software; it's central to the entire process, allowing businesses to tailor solutions that meet their exact needs while reducing reliance on expensive, one-size-fits-all software platforms.

As software becomes exponentially cheaper to build, we’ll see an influx of competitors vying for market share. While this explosion of options is great for innovation, it also means the days of monopolistic market dominance in software may be behind us. In a world where software can be built quickly and cheaply, the upside for any one company will be capped.

‍

The End of Traditional Software Metrics

Metrics like SaaS (Software-as-a-Service) and ARR (Annual Recurring Revenue) are built on the assumption that software is expensive to create and maintain. These metrics have served as a shorthand for evaluating a software company’s business model and its potential for growth. But as the cost of software creation plummets, these metrics will lose their relevance. The economic model that once fueled the software industry will no longer hold, and the market will be forced to correct itself.
This doesn’t mean the opportunity for massive venture returns is gone. On the contrary, we believe that solving deeply technical, difficult problems will still yield the next wave of 10,000x returns. However, the era of building simple “apps” with basic features and charging a monthly subscription fee is likely drawing to a close. In the post-AI world, those companies that can tackle complex, technical challenges will be the ones that capture the most value.

‍

The Path Forward

As we navigate this shift, the question becomes: how do we prepare for a world where AI is the primary driver of software development? At Bloomhaus Ventures, we believe the companies that will thrive in this new landscape are those that embrace AI, integrate it seamlessly into their development processes, and focus on solving the hardest problems.

We’re already seeing how LLMs are changing the game, and this is only the beginning. The future of software is one where AI takes on the heavy lifting, allowing human creativity and insight to focus on higher-order challenges. The companies that position themselves to take advantage of these shifts will not only survive—they’ll lead.

‍

Sources and further reading:

https://blog.hubspot.com/ai/klarna-ai-assistant

https://www.acquainted.studio/content/how-klarna-used-genai-to-revolutionise-its-customer-service

https://aiexpert.network/case-study-klarnas-revolutionary-use-of-ai-in-customer-service-and-operations/

https://www.crn.com/news/ai/2024/servicenow-ceo-bill-mcdermott-we-re-putting-ai-to-work-for-people

https://www.forbes.com/sites/quickerbettertech/2024/03/13/klarnas-new-ai-tool-does-the-work-of-700-customer-service-reps/

https://www.servicenow.com/content/dam/servicenow-assets/public/en-us/doc-type/resource-center/ebook/ebk-itsm-upgrade.pdf

https://finance.yahoo.com/news/servicenow-infusing-ai-everywhere-got-135718119.html?guccounter=1

https://finance.yahoo.com/news/amazon-ceo-andy-jassy-says-213018283.html

https://developers.slashdot.org/story/24/08/25/0049230/amazon-ceo-ai-assisted-code-transformation-saved-us-4500-years-of-developer-work

https://www.cnbctv18.com/technology/amazon-ceo-andy-jassy-says-gen-ai-saved-260-million-and-4500-developer-years-19465522.htm