AI Disrupts Software Value Beyond Code Generation

The "SaaSpocalypse" and the Looming Question of Software Value

The current market turmoil in the software sector, dubbed the "SaaSpocalypse," is fueled by a potent mix of AI-driven disruption and a growing realization that many software companies' valuations have been built on shaky foundations. While the immediate impact of AI on code generation might seem like a cost-saving benefit for software vendors, the deeper implication is a fundamental re-evaluation of what customers are actually paying for. This conversation with Jared Sleeper reveals that the true value of software lies not just in its code, but in the complex web of integrations, human expertise, and established ecosystems that are now facing unprecedented challenges. Investors and business leaders who grasp these hidden consequences and the long-term implications of AI will be best positioned to navigate this evolving landscape, identifying companies with durable competitive advantages rather than those built on ephemeral assumptions.

The Unseen Costs of Obvious Solutions: Why AI Isn't Just About Cheaper Code

The narrative surrounding the "SaaSpocalypse" often centers on the idea that AI, particularly advanced coding models, will drastically reduce the cost of software development, thereby devaluing existing software companies. While it's true that generating code is becoming significantly cheaper, this perspective misses the crucial, often hidden, costs and complexities that have historically underpinned the software industry's value proposition. Jared Sleeper, a seasoned software investor, argues that the true value of software has always been more than just the lines of code. It's embedded in the intricate integrations with legacy systems, the specialized human knowledge required for implementation and customization, and the "herd familiarity" that makes a product a de facto standard.

Consider the example of enterprise resource planning (ERP) systems. Sleeper points out that implementing such software is notoriously painful, often leading companies to miss earnings targets due to the sheer complexity of change management and integration. This complexity, he explains, is where much of the value--and stickiness--of software has resided. Companies pay not just for the software itself, but for the expensive consulting and internal effort required to make it work within their unique operational context.

The advent of AI, however, challenges this model. While AI can automate code generation, Sleeper suggests that the more significant human problem lies in understanding the existing systems and mapping them to new ones--a task that often requires deep institutional knowledge and human interaction. Yet, even this barrier is being eroded. He notes that for certain integrations, AI can now generate the necessary code, effectively solving problems that once required specialized engineering teams.

"It's not just software right now. So we're seeing this sort of rolling series of concerns where like every time AI does something or creates some new product, it hits a particular industry."

This shift forces a re-evaluation of what customers are truly buying. Sleeper introduces the concept of "herd familiarity," using Zoom as an example. Despite Microsoft Teams offering a free alternative, Zoom's ubiquity and user comfort create a valuable network effect. Similarly, the widespread knowledge of Microsoft Excel shortcuts makes switching to Google Sheets a non-trivial undertaking for many organizations. These are not just features; they are deeply ingrained aspects of how businesses operate. However, the core argument emerging is that AI is poised to automate not just the creation of software, but also the very processes that made it indispensable.

"Another thing that you sell when you sell as a software vendor is what I call herd familiarity, which means everyone on earth knows how to use your software, which just simplifies the training and onboarding workflow."

The existential threat, Sleeper elaborates, lies in the "terminal value concern"--the fear that in the long run, many of these software companies might be worth zero. This isn't necessarily about today's financial performance, which he notes has been surprisingly resilient, but about the forward-looking anxiety that AI will fundamentally alter the knowledge work landscape, potentially rendering entire categories of software and the roles they support obsolete. The rapid decline of Chegg following the rise of ChatGPT serves as a stark, real-time example of how markets can anticipate existential threats to software businesses, even before they fully manifest in financial results.

The Data Deluge and the Race for Contextual Intelligence

Beyond the code and the ecosystem, data is emerging as another critical battleground in the AI era. Sleeper highlights that while AI models may possess immense intelligence, their utility is severely limited without context. This context--understanding an organization's internal workings, data sources, and operational nuances--is precisely what many software companies, particularly those managing vast amounts of customer data, possess.

Salesforce, for instance, sits on a treasure trove of customer relationship data, including interaction history, sales pipelines, and support requests. Sleeper posits that an AI agent would desperately need access to this CRM data to be effective. However, he also points out a critical gap: much of the essential context resides not in structured data, but in human brains--the informal notes from a sales dinner, the understanding of a client's personal life, or the subtle nuances of a business relationship.

"It's an incredibly complex piece of software for a large enterprise. An AI agent working within a company, you would need access to that in order to get almost anything done, right? But you need more than is there."

This creates a race to capture this contextual information. Software companies have a foundational position, but AI-native startups are actively building agents designed to gather this missing context. The companies that successfully bridge this gap, becoming the "system of context" for AI agents, are likely to emerge as the most valuable in the new landscape. This suggests a future where software companies that can effectively integrate and leverage their proprietary data, augmented by AI, will thrive, while those that merely offer code generation or basic functionality will struggle.

The Shifting Sands of Software Value: From Tools to Outcomes

The conversation then pivots to a more nuanced view of software vulnerabilities and the potential for new value creation. Sleeper challenges the notion that all software is equally at risk, suggesting a spectrum of exposure. Companies serving enterprises with highly customized software or those requiring heavy implementation are, in his view, more vulnerable. This is because enterprises with significant resources and complex needs are precisely the ones most likely to leverage AI to replace core systems.

Conversely, small and medium-sized businesses (SMBs), like a family-run grocery store, are less likely to adopt radical technological shifts due to the pain and inertia associated with change. This implies that for certain segments, the traditional software model, with its emphasis on implementation and long-term adoption, might persist longer.

However, Sleeper also outlines a compelling bull case where software companies can not only survive but thrive by embracing AI. This isn't about simply adding more features to mature products. Instead, it's about a fundamental shift in how software is delivered and priced, moving from selling tools for employees to selling direct outcomes. He cites Intercom's "Finn" AI product as a prime example. By building an AI-native tool that directly answers customer support queries, Intercom has reaccelerated its growth, demonstrating a new paradigm of value delivery.

This transition necessitates a new pricing model--results-based pricing. Instead of per-seat licenses, companies can now sell outcomes, such as closed customer support tickets at a fixed price. This allows them to capture significantly more value, potentially offering customers a substantial return on investment even at higher price points. Sleeper illustrates this with the example of a sales rep earning $250,000-$300,000 annually. If a technology can effectively replace that rep, the software vendor could charge $50,000, still providing a five-fold ROI for the customer while dramatically increasing their own "take rate." This represents a potential "mega bull case" where software companies become orchestrators of AI-driven outcomes, akin to the transition from brick-and-mortar retail to e-commerce.

"And now we're just selling you an employee or the results of an employee. So, you know, we will sell you customer support tickets getting closed out for 50 cents or a dollar per ticket. And you can do the math of what it would cost you for the human to do that or what it would cost you for AI to do that. And we'll be cheaper, but we're also dramatically increasing what you pay us because, you know, we're cutting into a completely different stream."

This shift also raises questions about the core AI model providers. Sleeper acknowledges the risk that these foundation model vendors might move up the stack to capture more value, potentially commoditizing the resellers. However, he notes that many AI companies are already racing towards the application layer, understanding that user adoption and control over the application interface are crucial for long-term success and avoiding commoditization.

The Uncomfortable Truths: Layoffs, SBC, and the Future of Work

The discussion turns to the more uncomfortable realities facing the software sector, particularly concerning costs and workforce adjustments. Sleeper addresses the ongoing debate around stock-based compensation (SBC), a significant expense that is often excluded from non-GAAP reporting. He notes that while theoretically, AI could reduce the need for employees, many companies have been reluctant to cut headcount. This reluctance, he suggests, is partly due to management's reliance on non-GAAP metrics, which mask the true profitability.

"It creates this dynamic where, you know, yes, there's this terminal value concern, which by far the most important thing, but there's also no floor, right?"

Sleeper predicts that layoffs are inevitable. He argues that management teams will eventually respond to market signals and the need for financial discipline. Beyond cost savings, layoffs can also enable companies to move faster and retain top talent. The intense competition for skilled AI professionals means that companies must be able to offer competitive compensation, which may become more feasible after workforce reductions. Furthermore, he observes that employees who have not embraced AI tools are, paradoxically, becoming a drag on productivity, even a "negative marginal product."

The future of work, particularly for coders, is framed through an analogy to civil engineering. Just as civil engineers shifted from manual calculations to project management of computational tools, software engineers are increasingly becoming orchestrators of AI. While demand for software engineers is expected to remain strong in the near to medium term due to the vast undersupply of software globally, the nature of the work will evolve. Sleeper advises adaptability and a focus on uniquely human skills, such as sociability and relationship-building, which are harder for AI to replicate.

"And so I'm not necessarily bearish on the demand for software engineers, at least for the next, you know, three to five years. Beyond that, if things get weird, hard to tell."

Finally, Sleeper touches upon market dynamics, acknowledging the short-termism driven by hedge fund traders who cannot afford significant drawdowns. The lack of fundamental distress in many software companies makes it difficult to predict when the sell-off will end, creating a climate of fear. He also points to the absence of traditional valuation support mechanisms like buybacks and dividends, leaving many software stocks without a clear floor. The conversation concludes with a shared sense of uncertainty about how businesses will ultimately purchase and utilize software in the AI-driven future, underscoring the profound and potentially "weird" transformations ahead.

Key Action Items: Navigating the Software "SaaSpocalypse"

• Immediate Action (Next 1-3 Months):

  • Audit existing software stack for AI vulnerability: Identify core business software that relies heavily on proprietary code or complex integrations that AI could potentially replicate or bypass.
  • Assess internal AI adoption readiness: Evaluate current employee proficiency with AI tools and identify training needs to foster adaptability and prevent becoming a productivity drag.
  • Scrutinize non-GAAP reporting: For investors and finance teams, focus on GAAP profitability and free cash flow, looking beyond non-GAAP metrics that may obscure true financial health.

• Short-Term Investment (Next 3-6 Months):

  • Develop a data strategy for AI context: Begin cataloging and structuring proprietary data that can serve as critical context for AI agents, potentially enhancing existing software or creating new AI-native applications.
  • Explore results-based pricing models: For software vendors, investigate how to transition from per-seat licenses to outcome-based pricing that aligns with the value delivered by AI automation.
  • Invest in employee AI upskilling: Implement targeted training programs to equip employees with the skills to leverage AI tools effectively, fostering a more adaptable and productive workforce.

• Longer-Term Investment (6-18 Months and Beyond):

  • Build AI-native applications or features: For software companies, prioritize the development of AI-first products that offer direct outcomes rather than just tools for human users. This pays off in 12-18 months by creating new revenue streams and competitive moats.
  • Identify and cultivate human-centric roles: Focus on roles that require deep interpersonal skills, complex problem-solving, and relationship management--areas where AI is less likely to fully replace human capabilities. This creates durable advantage because these skills are inherently difficult to automate.
  • Strategic workforce planning for AI integration: Proactively plan for potential workforce adjustments, focusing on retaining top AI-proficient talent while managing the transition for roles that may be automated, understanding that this discomfort now creates lasting operational efficiency.