Will SaaS as We Know It Collapse in the Face of Agentic Applications?

This provocative question stems from a bold claim made by Satya Nadella during his BG2 podcast with Bill Gurley and Brad Gerstner . Nadella’s vision of the future centers around the evolution of an “AI layer” that could redefine how businesses operate and the role SaaS applications play. His three key assertions are nothing short of revolutionary:

1. Business logic will migrate into the AI layer.
2. The AI layer will simultaneously engage with multiple Systems of Record (e.g., Salesforce, Workday, NetSuite, SAP).
3. The database layer (System of Record) will become commoditized.

These statements have sparked widespread debate, but they also open the door to exploring the transformative potential of what Nadella describes as "agentic applications."

What Are Agentic Applications?

Agentic applications represent a new paradigm in business software, built on two key components:

1. LLMs (Large Language Models): These models provide essential capabilities for reasoning and planning, allowing them to process and understand complex business contexts.
2. Agents: These are components designed to take specific actions, such as making payments, sending emails, or tracking packages.

When combined, these elements (1 + 2) create a powerful framework for handling complex business processes and logic. Unlike today’s SaaS applications, which often rely on armies of implementation consultants to translate business needs into workflows, agentic applications promise a more seamless and efficient approach.

How Are SaaS Applications Built Today?

To appreciate the shift agentic applications represent, it’s important to understand how traditional SaaS applications operate. Using Salesforce CRM as an example, most SaaS platforms offer three primary components:

1. System of Record: At its core, this is a database—in Salesforce’s case, one that stores a list of customers and prospects.
2. UI for Input/Output: Actions performed in the user interface translate into one of the four CRUD actions (Create, Read, Update, Delete) applied to the System of Record.
3. Business Logic: Rules and workflows, such as automatically sending an invoice or notifying customer success when a deal is marked as closed, are pre-defined within the application.

These components have served businesses well but come with limitations, especially when it comes to flexibility, customization, and the cost of implementation.

Is SaaS Dead? Where Is the Opportunity?

Let’s address the elephant in the room: SaaS is not dead, but it’s on the brink of a major transformation. The rise of agentic applications does not mean the end of SaaS but rather an evolution. Here’s why:

1. Partial Automation: While AI will handle more complex logic and workflows, it’s unlikely that 100% automation will be feasible or desirable. Instead, the best solutions will integrate AI with human input, creating a seamless "human-in-the-loop" system.
2. Cost Reduction: One of the most significant opportunities lies in reducing the reliance on expensive SaaS implementations. Today, businesses spend millions on consultants to convert their needs into workflows. Agentic applications, powered by natural language processing and intelligent agents, promise to eliminate this inefficiency. Imagine defining your needs in plain English and having AI deliver the workflow instantly.

?? Imagine the acceleration when your Revenue Ops team can express a need in natural language and instantly create a workflow that would have taken weeks of Salesforce implementation. ??

The Challenge of Agentic and Future of Business Applications

Agentic applications herald a new era for enterprise software. By combining the reasoning and planning capabilities of LLMs with action-oriented agents, these applications offer a streamlined approach to managing business processes. They challenge the traditional SaaS model by:

• Simplifying implementation.
• Reducing costs.
• Allowing for greater flexibility and customization.

The Challenge of Agentic Applications lies technical limitations of LLMs such as incorrect reasoning, privacy concerns on sensitive data, and trust issues when it comes to errors. The best solutions will be the ones with a Human in the Loop

?? For SaaS business applications this means that while the database layer may become commoditized, the true value will lie in how the Agentic workflows naturally weave in the human-in-the loop interaction. This means notifications, approvals, and other interactions at the right steps of the flow.

Additionally quality will depend on how effectively the AI layer interacts with multiple systems of record and enables intelligent decision-making. For businesses, this means a shift from rigid, predefined workflows to dynamic, AI-driven processes—and ultimately, a more efficient and responsive way of working.

So, is SaaS as we know it on its way out? Not quite. But the emergence of agentic applications marks a pivotal moment. As businesses embrace this new paradigm, they’ll say goodbye to the inefficiencies of the past and hello to a future defined by innovation, agility, and AI-powered excellence.

?? The Nexla AI Integration platform offers the essential component for AI Layer -- agents that can execute tasks by seamlessly integrating into a variety of standard and home-grown enterprise systems.