AI will disrupt SaaS-But obsolescence is optional

For 25 years, SaaS transformed business by moving software to the cloud and delivering updates at digital speed. Generative and agent-based AI have now arrived to challenge that revolution with another - one where the software no longer merely serves the user, but completes the work itself.

Already, AI is drafting code, managing customer support workflows, and generating marketing content, shifting the paradigm from "user plus app" to "AI agent plus API." The velocity is unmistakable - OpenAI's o3 model dropped costs by 80% in just two months. Within three years, most repeatable, rules-based digital tasks could be handled entirely by AI.

The stakes for SaaS providers

Not all workflows will be disrupted equally. Mapping the terrain, there are four dominant scenarios:

1. Core Strongholds: Workflows with low automation potential and low AI penetration - think highly regulated tasks like clinical-trial randomization - remain safe harbors for incumbents. Here, AI should enhance, not replace, human expertise and unique data.

2. Open Doors: Tasks where AI can penetrate but human input is still valuable, such as list-building and task management, threaten incumbents with third-party AI agents siphoning value. Defending these requires aggressive adoption of agents and deepening integrations to lock in customers.

3. Growth Gold Mines: Where high automation is possible but market data is proprietary, incumbents can win by being first to deliver end-to-end AI automation - shifting business models from user-based pricing to outcome-based pricing.

4. Battlegrounds: Highly automatable workflows that are also wide open to AI penetration risk being commoditized or seized by the fastest-moving players. Only those who proactively rebuild their services for AI-first orchestration will thrive.

New layers, new rules

The traditional SaaS stack is fragmenting. Agentic AI is rebuilding control across:

• Systems of Record: The foundational data and business logic, where unique data gives a protective moat.

• Agent Operating Systems: Platforms orchestrating the actual work - think Azure AI Foundry or Google's Vertex AI Agent Builder.

• Outcome Interfaces: Simple, intuitive endpoints transforming requests into automated actions.

But the real bottleneck - and the next frontier - is the semantic layer: the syntax and vocabulary that lets agents truly speak the language of business tasks across platforms. The first mover to set this standard could become the marketplace that all others must join, setting new terms for influence and value creation.

What SaaS leaders must do

• Centre AI in your product roadmap. Automate repeatable jobs your software is already helping customers achieve.

• Protect and leverage unique data. Your proprietary information is the irreplaceable foundation that off-the-shelf models can't replicate.

• Standardize and open selectively. Define your data objects and workflows internally, then open the right elements to spur ecosystem adoption - without enabling fast-following competitors.

• Rethink pricing for an AI-first world. Start experimenting with outcome-based models as AI handles more of the work.

• Build AI fluency across every team. Make sure your organization - and your customers - understand and can articulate the value AI brings.

AI-driven disruption is non-negotiable, but obsolescence is not. The future of SaaS will be won by leaders who embrace AI as the engine of every workflow, rethink platform strategies, and write the next standard - before their competitors do.