Get your free personalized podcast brief

We scan new podcasts and send you the top 5 insights daily.

Companies no longer need SaaS products that simply analyze their data. They can now apply large language models directly to their data stores (like Salesforce or SAP) to generate insights, rendering an entire category of software obsolete and collapsing its pricing power.

Related Insights

The future of per-seat SaaS pricing is precarious. AI-driven productivity could shrink the number of knowledge workers, while LLMs can give casual users system access without a full license, eroding the user base from the periphery.

Enterprises no longer need to buy expensive SaaS products for tasks like customer feedback. They can now spin up custom AI agents internally, making it harder for SaaS companies to acquire new customers and leading to higher-than-modeled churn. This poses a fundamental threat to the SaaS business model.

SaaS tools whose primary value is aggregating and simplifying access to public information are vulnerable to being replaced by LLMs, which excel at this exact task. Defensible moats belong to platforms with proprietary data, deep workflow integration, and high regulatory barriers, not simple information convenience.

The threat to SaaS from AI isn't uniform. Foundational 'systems of record' like Salesforce are safe and being built upon with APIs. However, vertical SaaS tools (e.g., for surveys) are being replaced wholesale by custom AI-built solutions, justifying the concern over their declining growth and market caps.

Turing's CEO claims SaaS is dead for two reasons. First, powerful foundation models drastically lower the cost of building custom software internally. Second, existing SaaS products are built for human interaction via GUIs, not for AI agents that will increasingly use APIs and tool-calling functions directly.

The ease of building applications on top of powerful LLMs will lead companies to create their own custom software instead of buying third-party SaaS products. This shift, combined with the risk of foundation models moving up the stack, signals the end of the traditional SaaS era.

The accessibility of powerful LLMs has changed the competitive landscape for data analytics SaaS. Every product is now implicitly compared to a user setting up their own solution by pointing a model like Claude at their data warehouse. This forces SaaS companies to provide value beyond simple Q&A, like cost optimization and performance.

Wall Street believes AI is 'eating' software, causing stocks for giants like Salesforce and Oracle to plummet. AI tools like Anthropic's Claude Code, which can create software from simple prompts, threaten to undercut the value proposition of traditional Software-as-a-Service (SaaS) companies by democratizing and simplifying software creation.

SaaS products like Salesforce won't be easily ripped out. The real danger is that new AI agents will operate across all SaaS tools, becoming the primary user interface and capturing the next wave of value. This relegates existing SaaS platforms to a lower, less valuable infrastructure layer.

The key differentiator for SaaS companies is being "in the token flow," where AI model usage directly drives consumption of their product (e.g., more database queries). Companies outside this flow, like some front-facing apps, risk competing directly with AI models and face significant headwinds.