AI Agents Slash Software Valuations, Threaten SaaS Seat Model

AI agents are quickly replacing human users in software, causing big drops in software company values like Salesforce and Adobe. The old way of charging by the number of people using the software is breaking because now software can do the work itself. Wall Street is worried, and companies with lots of repetitive tasks are most at risk. Some software makers are trying to survive by focusing on data and making sure their tools are safe and useful for AI. The future may belong to those who help manage all the AI tools, not just those who sell software seats.

The rapid rise of AI agents is slashing software valuations and threatening the entire SaaS seat-based revenue model that has dominated the industry for two decades. As autonomous software becomes the primary user, investors are dumping shares in giants like Salesforce and Adobe, fearing the traditional per-seat pricing structure is now broken.

How AI Agents Squeeze Seat Licenses

AI agents threaten the SaaS model by bypassing graphical user interfaces (GUIs) to interact directly with backend APIs. This allows a single agent to perform tasks previously requiring multiple human users, drastically reducing the number of paid 'seats' that software vendors can monetize for revenue.

This fundamental shift is why Microsoft CEO Satya Nadella predicts that "business applications as we know them will collapse," as detailed in a Glean analysis. Instead of humans clicking buttons, autonomous agents now handle sales calls, draft proposals, and file expenses in parallel, shrinking the number of human logins that vendors can bill for. With McKinsey estimating that 80% of enterprise apps will embed agents by 2026, Wall Street models are now accounting for four key disruption paths:

Enhancing: Agents operate within existing tools, boosting productivity.
Compressing: A single agent stitches tasks across multiple apps, reducing subscriptions.
Outshining: Agents relegate SaaS tools to data backends, sidelining their UIs.
Cannibalizing: Unified AI layers replace entire software categories.

Bain finds that software for repetitive, structured-data tasks - like CRM, project management, and invoicing - faces the highest risk of compression and cannibalization.

Valuations in Freefall

The market reaction has been brutal. Doubts that new copilot features can offset lost seat revenue have driven significant selloffs, according to an InvestorPlace report. Over the last year, Adobe shares have shed over 24%, while Salesforce has lost 23%. During one severe session, Salesforce fell 6.5% and Adobe 5.4%, marking their steepest drops since 2024. Analysts now demand proof that new products like Salesforce's Agentforce and Adobe's Firefly can generate usage-based revenue before they will rerate the stocks.

A clear valuation divide is emerging. While UI-heavy vendors trade at depressed multiples, firms providing the AI orchestration backbone are attracting capital. Companies like C3 AI, UiPath, and IBM watsonx are positioning themselves as essential control planes that route tasks between models and data stores, a role that earns usage fees even without human logins.

The Pivot to AI Orchestration and Security

The market for AI orchestration is booming because while 65% of companies launch AI pilots, only 11% reach full deployment. Orchestration platforms bridge this gap. Prompts.ai, for example, aggregates over 35 large language models into a single dashboard with cost controls, while Palo Alto Networks' Cortex XSOAR automates security playbooks across 850 tools, turning agent proliferation into a revenue source.

Security is a top priority, as Valence Security warns of "two orders of magnitude more risk" from agents exchanging credentials. Governance features like audit trails, fine-grained permissions, and outcome tracing are becoming critical purchasing factors, rivaling raw model accuracy.

The Path Forward: Data Stewardship or Obsolescence

Investors are closely tracking the shift from