AI Agents Dictate Software Adoption, Threatening Human-Centric Models

The AI Agent Revolution: Why Traditional Software Models Are Facing an Existential Threat

This conversation reveals a seismic shift in the software landscape, driven by the ascendance of AI agents. The core thesis is that the future of software adoption will be dictated not by human preference, but by the autonomous decisions of AI agents. This has profound, non-obvious implications: traditional software companies that rely on human workflows risk obsolescence, while those that integrate seamlessly with AI agents stand to gain immense competitive advantage. Founders, investors, and enterprise leaders who grasp this transition now will be best positioned to navigate the coming disruption, identifying durable businesses and avoiding those destined for obsolescence. Ignoring this fundamental change means betting on a past that is rapidly disappearing.

The Agentic Mandate: Why Human-Led Workflows Are Becoming Obsolete

The conversation highlights a critical disconnect between how the market currently perceives AI's impact and the reality of its trajectory. While much attention is focused on foundation models like OpenAI and Anthropic, the true battleground is the "agentic layer"--the AI agents that will increasingly orchestrate workflows. Jason argues that agents, unlike humans, will not be swayed by brand loyalty or familiarity. Their choices will be purely utilitarian, prioritizing efficiency and efficacy. This means that software previously indispensable for human productivity, such as project management tools or even design platforms like Canva, could become irrelevant if agents can generate superior outcomes directly. The implication is that companies built around human-centric workflows are fundamentally vulnerable.

"The agents are going to choose which LLM we use just like everyone from dario down has said it's going to be more and more agents doing coding the agents are going to make the decision on everything."

This shift fundamentally redefines competitive advantage. For years, the narrative has been about building better models or securing more compute. However, the real advantage will lie in becoming the preferred platform for AI agents. Companies that can demonstrate that their tools are essential for agents--providing unique data, APIs, or capabilities that agents can leverage--will thrive. Conversely, those that remain human-centric risk being bypassed entirely. The conversation touches on how even market leaders like Marketo and Outreach are viewed as potentially useless by agents, as agents can craft and send superior emails directly. This suggests a future where the value proposition for enterprise software shifts from facilitating human tasks to enabling agentic operations.

The Compute-to-Revenue Correlation: A Risky Bet on Infinite Demand

The massive investments in Anthropic, while seemingly a validation of AI's potential, also underscore the immense capital intensity and inherent risks in the compute-to-revenue model. As Rory and Jason discuss, building AI models is only half the battle; securing and scaling the necessary compute is a monumental and capital-intensive undertaking. The assumption that compute directly correlates to revenue, a belief championed by figures like Sam Altman, is being tested. While compute is a prerequisite for revenue ("no compute equals no revenue"), a great model is also required to generate demand.

The sheer scale of investment needed to meet projected demand is staggering. For a company like Anthropic, projecting a five-fold increase in revenue within two years implies needing to invest hundreds of billions of dollars in compute infrastructure. This capital expenditure must be made years in advance of the revenue it's intended to support, creating a significant risk of stranded assets if demand forecasts are wrong.

"You're doing 10 billion in run rate and you're effectively saying between you and your partners to be able to meet demand two years from now you've got to invest 300 billion not all yourself some but your partners but think how and by the way if you get it wrong and you end up doing 200 billion in run rate revenue you're going to have half the compute you need you're going to look like an idiot and if you get it wrong and you only get to 50 billion in revenue two years from now you're going to be left with 150 billion of stranded capacity."

This capital intensity is a stark contrast to the software industry's historical advantage of low marginal costs. The risk is amplified by the two-year lead time for securing compute, forcing companies to make massive bets on future demand. This dynamic creates a scenario where companies can face periods of both insufficient and excessive compute, leading to significant short-term volatility. The conversation suggests that while the long-term trend of increasing compute demand driven by agents is likely, the short-term swings and the capital risk are substantial.

The Great Software Consolidation: When Debt Becomes a Death Sentence

The discussion around Thoma Bravo handing over Medallia to creditors is a stark illustration of how the evolving software landscape, particularly the rise of AI and the increasing cost of capital, is exposing vulnerabilities in traditional private equity models. Medallia, a company in the customer engagement space, faced a perfect storm: a business model that is not inherently AI-first, a need for significant investment to pivot, and substantial debt.

The key takeaway is that overpaying for companies, even those in seemingly stable sectors like enterprise software, can be fatal when combined with debt and a lack of a compelling AI narrative. Medallia's debt load, while not excessively high relative to its historical valuation, became unsustainable on its current revenue and growth profile, especially given its pre-AI story.

"You can't service two billion plus of debt on a one billion low growth company with a pre ai story that has to transform to ai you simply can't."

This situation is not isolated. The conversation points to a cohort of similar companies--New Relic, Anaplan, Zendesk, and others--facing similar pressures. The rise of AI-first alternatives in areas like customer sentiment analysis means that traditional software companies may struggle to retain relevance, let alone grow. This creates a "melting iceberg" scenario where even without extreme leverage, the lack of an AI story and the difficulty of a full rewrite make them unattractive. The implication for investors and operators is that the exit landscape is narrowing. IPOs require massive scale and growth, strategic acquirers are highly selective, and private equity, once a reliable buyer, is now facing its own challenges with debt and the need for AI-enhanced businesses. This forces a rationalization, where companies that cannot adapt may be consolidated or simply given back to creditors, signaling a significant shift in the dynamics of the software market.

Key Action Items

• Immediate Actions (Next 1-3 Months):

  • Re-evaluate software stack for agentic compatibility: Identify which tools in your current stack are human-centric and explore AI-native alternatives or integrations that agents can directly leverage.
  • Assess compute strategy: Understand your current compute utilization and projected needs, considering the capital intensity and risk associated with scaling compute infrastructure.
  • Scrutinize debt structures: For companies with significant debt, analyze their AI narrative and growth potential. If the AI story is weak, begin planning for potential restructuring or deleveraging.
  • Prioritize AI integration over human workflow optimization: Shift development resources towards building capabilities that agents can utilize, rather than solely enhancing human user interfaces.
  • Analyze vendor consolidation impact: Understand how AI-driven consolidation trends might affect your product's defensibility and market position.

• Longer-Term Investments (6-18 Months & Beyond):

  • Develop an AI agent strategy: Define how your company will engage with, serve, or become an essential platform for AI agents. This is a strategic imperative, not an optional add-on.
  • Invest in proprietary data or unique APIs: Build assets that agents will need to access to perform their tasks effectively, creating a moat that is difficult for competitors to replicate.
  • Explore strategic partnerships with foundation model providers: Understand how your product can integrate into the agentic ecosystems being built by major AI players.
  • Focus on durable revenue streams: Prioritize business models that are resilient to agentic disruption, potentially through essential data provision or unique computational capabilities.
  • Prepare for a narrowed exit landscape: Understand that traditional exit paths (IPO, PE sale) are becoming more challenging and require significant scale and AI relevance. Plan for fewer, but potentially larger, outcomes.
  • Embrace immediate discomfort for future advantage: Invest in AI-native architecture and agent integration now, even if it requires significant upfront effort and no immediate visible payoff. This discomfort will create a lasting advantage as the market shifts.