SaaS Isn't Dead Yet, But It Certainly Is Doomed
1
12
0
As Agentic Workflows of AI Agents take over operational tasks, the foundation of Software-as-a-Service is rapidly becoming outdated. In this article, I outline why this shift is more than technological - it’s a transformation of the entire commercial model for enterprise software.
-------------------------------
Executive Takeaways:
Leaders must adopt a new kind of systems thinking, one rooted in dynamic orchestration, not static interfaces. This shift demands robust technical governance to ensure operational efficiency, security, compliance, and control across a distributed and increasingly autonomous ecosystem.
Leaders must rethink financial governance in many areas. Business case development, procurement, usage metering, workforce planning, and accounting practices to align workforce productivity with a FaaS model instead of a SaaS one.
-------------------------------
Why Is SaaS Dying?
Everyone’s still buzzing about Satya Nadella's bold claim that “SaaS is dead.”
Most responses have zeroed in on technical shifts: new app architectures and changing roles for developers. Sure, that’s certainly a major part of the story. But those aren’t the reasons SaaS is truly facing extinction. The deeper disruption is something far more significant for today’s business models.
But before we unpack the real reason SaaS is on the chopping block, let’s ground this in something real. I’ll walk you through a practical enterprise example, where AI Agents quietly take over a small part of a human’s day job. Shout out to Brij kishore Pandey, Avi Chawla, Deepak Bhardwaj, and Rakesh Gohel, your insightful technical visuals are outstanding (and please keep them coming!), but for this one, we need a real-world use case explanation.
AI Agents Significantly Change How Work Is Done
The Agentic Workflow illustrated here and documented below showcases an Agentic AI-driven process helping Denise complete HR onboarding tasks. This isn’t theory, it’s a practical glimpse into how AI Agents are stepping into real operational roles.
Denise is a seasoned HR case consultant with deep expertise in employment regulations across multiple countries. Like many professionals, she used to spend a large portion of her day doing manual tasks, such as setting up payroll for new hires. While her company had already introduced some automation (hardcoded script workflows, highlighted in bold in the documentation at bottom), Denise still had to gather offer letter details, apply country-specific rules like tax deductions and state-specific leave entitlements, and feed that data into the system herself using the tools.
Now, everything is shifting. With AI Agents entering the scene, Denise no longer needs to manually chase down this information. She simply triggers the AI Agent orchestration, by clicking a button, saying a command, or selecting an option, and the system takes over. The Agentic Workflow reads the offer letter, interprets the role classification, applies local employment rules, and compiles everything needed for payroll submission. The heavy lifting is now handled autonomously, allowing Denise to focus on higher-value work.
Moving From Supervised to Autonomous
As shown in the illustration right now, the final actions (writing all the final data into the payroll system, which is shown in light red text) is still in supervised mode. The AI Agents do that heavy lifting for Denise, assembling all the necessary details, but she (a human) still reviews and approves the data before it's officially submitted to the system of record.
But that’s only a temporary phase. Once the full Agentic Workflow consistently matches Denise’s level of accuracy, this checkpoint can be removed. The process shifts from supervised to fully autonomous, with the AI Agents not only gathering and preparing the data but also submitting it – no human required. This is the real change: from task automation to decision automation.
Now, imagine this pattern applied across all the tasks of all the case types that an HR case agent handles today, whether manually or through basic scripted workflows. The implications aren’t just incremental, they’re transformational.
Five Major Considerations of AI Agents
First, if Denise is no longer performing hands-on HR casework, will her job role stay the same? Not at all. Her value now shifts. Denise can transition into a more strategic role, as a domain expert overseeing the collaboration with an AI Agent developer like Bob. Together, they ensure the AI Agents are designed and orchestrated to reflect the correct business logic, uphold compliance with legislative requirements, apply correct locale values, and incorporate updates to the payroll onboarding process as laws or policies evolve. In this model, Denise’s expertise isn’t lost, it’s elevated.
Second, this shift from human-led tasks to AI Agent execution isn’t limited to HR, it’s universal. Every industry, every job function is in scope. Any action a human performs on a computer, whether it’s filling forms, making decisions, or interacting with digital systems, can be encoded in LLMs or as a script, API call, workflow, search retrieval, or RPA routine. This is not just a trend in enterprise software. It’s the future of work. Remember Bill Gates's remarks on doctors and teachers? This transformation is coming for all professions, even those who may think that they’re outside the “productivity worker” bubble. Spoiler: they’re not.
Third, as decision-making and business logic shifts from humans to AI Agents, we’re witnessing a fundamental architectural transformation. Logic is no longer housed in front-end user interfaces designed for human cognition. Instead, it’s being executed through algorithms, via LLMs, machine learning models, and autonomous workflows. Once the core business logic moves to the AI Agent tier, there’s little value in building traditional, task-based UIs for human users. The interface now belongs to developers like Bob, who work with experts like Denise to encode reasoning into agent orchestration. From there, the Bobs of the world use their developer tools to build, deploy, and manage a fleet (or swarm) of AI Agents, each executing business logic independently. A shift from user-facing interaction to machine-executed logic at scale.
Fourth, AI Agents won’t be tied to any single platform or database. As Satya Nadella highlighted, these agents will operate across multiple data repositories, executing Create, Read, Update, and Delete (CRUD) operations not just within their local system, but across any backend they can access via integration. Whether it’s a prebuilt connector or a custom API, it doesn’t matter. AI Agents are data-agnostic. They don’t care where the information lives, they just need access to it to get the job done.
Fifth, the competitive edge in business software is shifting, fast. We’re moving from traditional human-facing productivity tools to advanced “super-AI” platforms built to manage fleets of AI Agents. With technologies like MCV and A2A orchestration enabling shared memory technology units and cross-platform collaboration respectively, these AI management systems won’t be tied to any single vendor ecosystem. The real workforce value will no longer come from those skilled in “point-and-click” software. It will come from those who can design, instruct, and continuously refine AI Agents. This is the true meaning behind the soundbite “AI won’t take your job, but someone who knows how to use AI will.”
Here Is Why SaaS Is Dying
So, what does all this mean for SaaS, and why might we soon declare it dead?
At its core, SaaS is not a technology model – it’s a commercial one. It’s based on a straightforward equation:
(X units) × (Y dollars per unit per month) × (12 months) = Annual Contract Value (ACV).
In nearly every case, the "X" in that equation is a human being. Someone logging into a productivity application to perform task-specific work.
But as we’ve seen in points 1 through 5, that foundational assumption is breaking. Humans like Denise are no longer the primary actors driving work through these systems. As Satya Nadella pointed out, the business logic is shifting to the AI Agent tier, meaning people are no longer logging in to get work done. The agents are doing it on their behalf.
This collapse of the human-as-user model means SaaS, as a unit of commerce, loses relevance. Just like infrastructure moved to a metered, pay-as-you-go utility model through providers like AWS, Azure, and GCP, business productivity is following suit. It’s becoming a computational workload, executed, measured, and billed per task or outcome, not per person. Consumption and usage is the emerging New Normal.
Forward-looking vendors already see it coming. As AI Agents replace human users, the old annuity models of seats, CALs, and app users are beginning to erode. In response, new pricing models are emerging: tokens, credits, assists. All designed to monetize based on functions, not users.
This is more than just a pricing shift; it’s a complete rewrite of how software value is delivered and captured in an AI-first world.
This is why SaaS is dying. The logic layer is no longer human. The unit of measure is no longer a seat. The age of business productivity Function-as-a-Service (FaaS) has begun.
Two Critical Takeaways For Business Leaders
First, organisations must begin understanding the new systems architecture that AI Agents are rapidly introducing. The era of “point-click-type” productivity tools is being replaced by a paradigm of designing, instructing, and improving AI Agents to execute work. Leaders must adopt a new kind of systems thinking, one rooted in dynamic orchestration, not static interfaces. This shift demands robust technical governance to ensure operational efficiency, security, compliance, and control across a distributed and increasingly autonomous ecosystem. (An excellent new article is here by Piyush Ranjan: MCV vs. Google’s A2A Protocol: Two Diverging Paths in the Evolution of Agentic AI.)
Second, financial governance must evolve just as quickly. The traditional models of cost attribution based on "human seats” of licenses are becoming obsolete. As AI Agents take over task execution, productivity will become measured in functions performed, not people employed. That means leaders must rethink financial governance in many areas: business case development, procurement, usage metering, workforce planning, and accounting practices to align workforce productivity with a FaaS model instead of a SaaS one.
This is not a distant horizon – it’s already underway.
The organisations that act now, those who adapt their governance frameworks today, will be the ones best equipped to lead tomorrow.
Use Case: Payroll and Compliance
Description: This use case automates payroll and compliance tasks for new hires using multiple AI agents. It determines employment type, assigns bonus plans, sets up salary structure, configures tax withholdings, deductions, and benefits enrolment, handles payroll setup reminders, monitors compliance, and creates cases for discrepancies.
Instructions (prompt to the LLM):
1. Fetch employment type details from user details.
2. Fetch Offer Letter details and save in memory to use later.
3. Set up salary structure based on offer letter details.
4. Fetch and assign additional bonus plans based on the employment type.
5. Configure tax withholdings based on the employee’s geographic location.
6. Configure deductions based on the employee’s geographic location.
7. Configure benefits enrolment based on the employee’s geographic location.
8. Summarize the complete activities performed and add the task work notes.
Note: Once completed, check with the agent to see if needs anything else on the case.
-- There are eight (8) aligned AI Agents that the AI Agent Orchestrator (coordinating LLM) can call on to perform actions --
(1) AI Agent: Geo Location Deduction Agent
Description: This AI Agent is designed to automate the process of configuring deductions based on an employee’s geographic location. It is intended for use by HR and payroll teams who need to ensure that deductions are accurately calculated and applied based on location-specific policies. The agent’s purpose is to improve payroll accuracy and reduce manual effort.
AI Agent Role (prompt to the LLM): You are responsible for automatically configuring deductions based on an employee’s geographic location. You will analyze the employee’s location data and apply the appropriate deduction policies accordingly. You will ensure that deductions are accurately calculated and applied, reducing manual effort and improving overall payroll accuracy.
Instructions (prompt to the LLM):
1. Analyze employee location data to determine deductions.
2. Ensure accurate calculation and application of deductions.
3. Apply deductions to the payroll.
4. Improve overall payroll accuracy.
Tools (available for use by the AI Agent LLM):
- Fetch Deductions Based on Location (Autonomous Script)
- Update Deductions to Payroll (Supervised Script)
(2) AI Agent: Offer Letter Parser Agent
Description: This agent is designed to assist HR teams by automating the process of extracting key information from offer letters. It can help streamline onboarding processes and ensure accurate data capture for payroll and other systems.
AI Agent Role (prompt to the LLM): You are responsible for extracting payroll and other necessary details from offer letters. You will analyze the offer letter text to identify and retrieve relevant information such as salary, benefits, start date, and other employment terms.
Instructions (prompt to the LLM):
1. Ask, “Would you like me to proceed with fetching the offer letter details?”
2. Once the agent confirms, Extract payroll details from the offer letter text.
3. Retrieve other necessary details such as start date, benefits, and employment terms.
4. Analyze offer letter text to identify relevant information.
5. Ensure accurate data capture for payroll and other systems.
6. Assist HR teams in streamlining onboarding processes.
Tools (available for use by the AI Agent LLM):
- Fetch Offer Letter Details (Autonomous Script)
(3) AI Agent: HR Case User Details Fetcher
Description: This AI Agent is designed to assist in HR case management by fetching user and HR profile details of the subject person field. It is intended for HR personnel and support agents who need quick access to relevant information for case resolution.
AI Agent Role (prompt to the LLM): You are responsible for fetching user and HR profile details of the subject person field in the HR case. You will retrieve relevant information from the HR case and present it to the user, aiding in efficient case management and resolution.
Instructions (prompt to the LLM):
1. Fetch user details based on the subject person field in the HR case.
2. Fetch Employment type information.
Tools (available for use by the AI Agent LLM):
- Fetch User Details (Autonomous Script)
- Fetch Case Details (Autonomous Script)
(4) AI Agent: Update Work Notes or Additional Comments
Description: Update Work Notes or Additional Comments on the HR Case record.
AI Agent Role (prompt to the LLM): You are to take input of the message and update the work notes or additional comments on the case record.
Instructions (prompt to the LLM):
1. Get details like message, number, etc.
2. Pass to the Update Work Notes or Additional Comments tool.
Tools (available for use by the AI Agent LLM):
- Update Additional Comments (Autonomous Script)
- Update Work Notes (Autonomous Script)
(5) AI Agent: Tax Withholding Agent
Description: The Tax Withholding Agent is designed to automate the process of configuring tax withholdings based on employee location. It is intended for use by HR and payroll departments to ensure accurate and compliant tax withholdings for all employees. The agent will use geolocation data to determine the appropriate tax rates and apply them automatically, reducing the need for manual intervention and minimizing the risk of errors.
AI Agent Role (prompt to the LLM): You are responsible for automatically configuring tax withholdings based on the geographic location of employees. You will analyze the employee’s location data and apply the appropriate tax withholding rates accordingly. You will ensure compliance with tax regulations and provide accurate and up-to-date tax withholdings for each employee.
Instructions (prompt to the LLM):
1. Analyze employee’s geographic location data.
2. Determine appropriate tax withholding rates based on location.
3. Apply tax withholding rates to the employee’s payroll.
4. Ensure compliance with tax regulations.
5. Provide accurate and up-to-date tax withholdings for each employee.
Tools (available for use by the AI Agent LLM):
- Fetch Tax Withholding Rates Base on Location (Autonomous Script)
- Update Tax Withholdings Rates to Payroll (Supervised Script)
(6) AI Agent: Salary Setup Agent
Description: The Salary Setup Agent is designed to automate the process of setting up salary structures based on offer letter details, making it easier and more efficient for HR personnel to manage employee compensation. It is intended for use by HR teams and managers who need to set up or update salary structures for new hires.
AI Agent Role (prompt to the LLM): You are responsible for setting up the salary structure based on the details provided in the offer letter. You will analyze the offer letter, extract relevant information such as salary, benefits, and other compensation details, and then create or update the corresponding salary structure in the system.
Instructions (prompt to the LLM):
1. Analyze offer letter details available to extract salary information and display information to the user.
2. Validate with the agent if any changes are required.
3. Once confirmed, set up the salary structure for the new hire.
4. Communicate any issues or discrepancies to HR personnel for review and resolution.
Tools (available for use by the AI Agent LLM):
- Salary Setup Structure (Supervised Script)
(7) AI Agent: Bonus Plan Assigner Agent
Description: The Bonus Plan Assigner Agent is designed to automate the process of assigning bonus plans based on employment type. It is intended for use by HR personnel and managers who need to assign bonus plans efficiently and accurately.
AI Agent Role (prompt to the LLM): You are responsible for assigning bonus plans based on the employment type of the user. You will validate the employment type and assign the appropriate bonus plan accordingly.
Instructions (prompt to the LLM):
1. Assign an appropriate bonus plan based on the validated employment type.
2. Ensure accuracy and efficiency in assigning bonus plans.
Tools (available for use by the AI Agent LLM):
- Fetch Bonus Plans (Autonomous Script)
- Assign Bonus Plan (Supervised Script)
(8) AI Agent: Geo Benefits Enrollment Agent
Description: The Geo Benefits Enrollment Agent is designed to streamline the benefits enrolment process for employees by automating the configuration based on their geographic location. It is intended for use by HR personnel and employees to ensure accurate and compliant benefits enrollment.
AI Agent Role (prompt to the LLM): You are responsible for automatically configuring employee benefits based on their geographic location. You will validate the employee’s location and apply the corresponding benefits package, ensuring compliance with regional regulations and company policies.
Instructions (prompt to the LLM):
1. Validate employee’s geographic location.
2. Apply the corresponding benefits package based on location.
3. Ensure compliance with regional regulations and company policies.
4. Communicate configuration details to HR personnel and employees.
Tools (available for use by the AI Agent LLM):
- Fetch Benefits Based on Location (Autonomous Script)
- Update Benefits to Payroll (Supervised Script)
