Is AI going to kill SaaS? The rise of AI software agents

Over the last few years, there has been a lot of talk about AI killing everything.

AI is killing writing. It is killing customer service. It is killing programming. At this point, I think the only thing AI hasn't been accused of killing is coffee breaks.

There's one industry where the conversation gets particularly interesting though: SaaS (Software as a Service).

Ordinarily, you have to guide software through each step to get the desired result. AI software agents are now changing that by handling more of the steps.

And that raises a question many people shy from: If AI agents can use software on our behalf, isn't AI killing SaaS?

This article answers the question and shares deep personal thoughts about the future of SaaS.

What are AI software agents and how are they different from traditional software?

Before we start predicting the death of SaaS, let's get one thing straight: an AI software agent isn't simply an AI assistant that helps you write an email or summarize a document. It goes a step further.

An AI agent software is a program that's capable of performing tasks on behalf of a user without following a fixed set of instructions. Salesforce describes this as the difference between traditional software that follows a predefined script and an AI agent that can decide what to do and which tools to use.

Think about a salesperson trying to find new customers.

With traditional software, the person opens a CRM, searches for prospects, looks up information about their companies, decides which ones are worth contacting, writes a customized email for each prospect, sends them, and then updates the CRM.

It's different with AI agent software. All the person has to do is give the AI agent an instruction, such as:

"Find promising companies in this market, research the right contacts, prioritize the strongest opportunities, prepare personalized emails, send them, and document their updated info."

How AI software agents are reshaping the SaaS business model

Software AI agents are reshaping how people interact with SaaS, how companies charge for it, and even how businesses decide which software product to buy.

Software interface will become less important

Since when SaaS became a thing, the interface has been a significant part of how people interact with the software. You even see some brands using the seamlessness of their software interface as their unique selling point.

Rightly so, but that is changing as AI-powered software agents are growing rapidly. There are already notable people in the industry making this argument.

Bret Taylor, the former co-CEO of Salesforce and the current co-founder & CEO of Sierra, in an interview with TechCrunch, said that software is moving away from the traditional model of clicking through applications. He argued that people will soon use language-driven prompts to complete tasks without clicking buttons on the software interface.

Top SaaS companies are rethinking per-seat pricing

SaaS per-seat pricing model is built around counting the number of people using the software. If a company has 10 employees who need access, they buy 10 seats. If the company grows to 50 employees, they buy 40 more seats.

But AI agents in SaaS don't operate like that. An agent isn't an employee. It doesn't need an email address or a login the same way a person does.

So imagine a company with 50 salespeople. They have an AI agent handling the prospect research, updating records, preparing outreach emails, and repetitive follow-ups.

Why should the company continue paying for 50 seats?

Top SaaS companies are already experimenting with this. For example, Salesforce's Agentforce pricing includes a consumption-based option in which customers pay according to the actions their agents perform.

HubSpot is also experimenting with a similar approach. Alongside its subscription plans, some AI agents consume credits to perform their functions.

Change in how businesses discover and buy SaaS

Today, buying SaaS usually involves searching Google, opening several websites to read reviews, comparing features, checking pricing, and then using a free trial.

Now imagine giving the entire research job to an AI agent. You could tell it to:

"Find a project management tool for our 30-person remote team. It needs to integrate with Slack, support automated workflows, and cost less than $500 a month."

Instead of giving you a list of links and asking you to do the research yourself, AI agents for business will compare several project management tools against those requirements and narrow down the options.

HubSpot, Salesforce, and Intercom show what AI software agents are already doing

You don't have to rely on my personal conviction to understand where this is going. Some of the biggest names in SaaS are already building AI agents into their products.

HubSpot is turning AI into more than an assistant

HubSpot has several AI agents, including the Prospecting Agent, which can research accounts, identify contacts, find buying signals, and create personalized outreach. Its Customer Agent, which can handle customer questions, and its Data Agent works with business information to provide answers.

Salesforce wants agents to become part of the workforce

Salesforce has a platform called Agentforce. It's for building and deploying AI agents that can work with business data and use tools to complete tasks. Those agents can support areas like sales, customer service, marketing, and commerce.

Intercom shows how far this can go in customer service

Intercom's Fin AI agent is designed to resolve customer questions rather than simply give a support representative a suggested answer. It can use a company's knowledge and information to handle conversations with customers.

So, is AI actually going to kill SaaS?

No. At least not in the way people usually mean when they say AI is going to kill SaaS.

AI software agents aren't making software unnecessary. In fact, they're making software more useful by giving it the ability to do more without waiting for a person to direct every step.

But that doesn't mean SaaS applications can carry on exactly as they have for the last decade. Some parts of the traditional model are clearly under pressure.

SaaS products that exist mainly to help people complete repetitive tasks will become less valuable when an AI agent can handle those tasks.

Interfaces built around endless clicking and form-filling will matter less when users can simply tell an AI software agent what they want.

Per-seat pricing will become harder to justify when an AI agent, rather than a human employee, is doing much of the work.

And software discovery will change if businesses increasingly use AI to research and compare products.

What SaaS companies need to do before the AI agent shift accelerates

If you're thinking of the future of SaaS, you need to think beyond simply adding a chatbot to a product and calling it an AI agent. Do these things instead:

Build products that AI software agents can work with

An AI agent needs a way to communicate with software. That's where APIs become important. They provide a way for applications to exchange information and trigger actions without a person manually moving data between them.

Focus on outcomes, not just features

AI agents in SaaS don't care whether your product has 100 features. What they are concerned about is whether your product can help achieve a particular task. The clearer the answer is, the easier it is for your product to be recommended.

Rethink pricing model

Per-seat pricing made sense when people were the primary users. But if software AI agents begin to perform more of the work, you should evaluate whether per-seat pricing still matches the value users receive.

Make your product easy for AI to understand

SaaS companies have spent years making websites attractive to humans and search engines. Now, you must optimize your software to be cited by LLMs. That means you should provide detailed product information, accurate pricing, useful reviews, and strong third-party mentions.

Keep humans in control

Many people mistake AI-powered software agents for autonomous software. The more responsibility you give the agent, the more carefully you need to control what it can do. You need to tell it what it's permissible to do on its own and when a human command is needed.

FAQs

What are AI software agents?

AI software agents are AI-powered systems that can understand a goal, determine the steps needed to achieve it, use available tools, and take actions with less human intervention.

How do software AI agents work?

A software AI agent receives an objective, considers the information and tools available to it, decides what actions are needed, and carries out those actions to reach the goal.

What is the difference between AI agents and chatbots?

A chatbot primarily responds to conversations. An AI agent can go further by taking actions such as updating the CRM and preparing outreach based on the conversation had.

Will AI software agents replace SaaS?

Not entirely. AI software agents still need applications, data, infrastructure, security controls, and record systems to perform reasonable work.

What can AI software agents do for businesses?

It depends on the software and permissions granted. AI agents for business can perform tasks such as researching prospects, analyzing data, updating records, and transferring data between applications.

Conclusion

AI software agents are proposing a different deal than what we're accustomed to: Tell the software what you want, and it will figure out how to accomplish it.

Does that mean AI is going to kill SaaS? No, but it will change how we traditionally expect software to work. The biggest SaaS companies, such as HubSpot, Salesforce, and Intercom, are already moving in this direction.