The main driver for centralizing data is shifting from business intelligence to providing essential context for AI agents. Without a unified data source, agents are as limited as pre-internet ChatGPT, unable to understand current business realities.
The primary danger for established SaaS companies isn't that AI agents will replace their UIs. The larger threat is that AI-native startups can now build superior products so quickly that they can rapidly catch up to and overtake incumbents.
Businesses are unlikely to use powerful AI simply to shave a few percentage points off their software spend. The real, high-impact ROI comes from applying AI to improve core business operations, making the actual business more effective and efficient.
Incumbent SaaS companies are starting to block API access for AI agents. They fear agents will bypass their user interfaces to perform the same functions, devaluing their core product and eroding the traditional per-seat revenue model.
To combat vendors restricting data access in the AI era, customers with significant purchasing power should proactively write language guaranteeing data access rights into their Master Service Agreements (MSAs). This creates leverage and sends a clear signal to the market.
The concept that data is too large and costly to move is an illusion created by legacy pipelines that repeatedly copy entire datasets. Fivetran's CEO asserts that modern change-data-capture techniques make data movement small-scale and inexpensive.
Treating AI agents as individual users who join Slack and get onboarded by HR is an effective, transitional strategy because it fits into existing human-centric workflows. The end state, however, is likely a more efficient, closed-loop AI system with a unified identity.
Despite predictions of SaaS's collapse, leading AI companies like OpenAI and Anthropic are still significant customers of traditional SaaS tools. This suggests that AI agents are augmenting, not completely replacing, established enterprise software.
Top AI labs like OpenAI and Anthropic build internal data platforms with conventional tools like Fivetran and Snowflake. This indicates a modern data stack is perfectly sufficient for providing AI context, and companies don't need to build bespoke, exotic infrastructure.
Contrary to its popularity, Postgres is old technology with significant technical debt. The CEO argues that AI coding agents make it feasible to build a superior, modern operational database from scratch, breaking the industry's reliance on legacy systems.