What are AI agents and are they the future of SaaS?

What are AI agents, really?

AI agents (at least right now) are essentially automation workflows with some AI bots making decisions along the way.

Example: I have a Zapier automation which triggers every time someone buys one of my courses on Lemon Squeezy and invites them to the course on Podia where I host the content.

It works perfectly every time doing this very simple task.

But what if I wanted to automate something where the flow is more ambiguous?

Example: Imagine I’m selling a product over email where in some situations I get questions from potential buyers? If I wanted to automate this, a simple automation wouldn’t work. Perhaps I could put in a filter to search for question marks in the email, but that’s pretty hacky!

This is where agents are useful. By putting AI into the automation flow, the AI could make intelligent decisions on whether the buyer is asking a question. It could then alter the path of the automation, not based on a simplistic condition, but based on an understanding of the context of the task and its goals.

It’s the equivalent of swapping a summer intern who does only what they’re told for a 23-year-old graduate who understands context and has some autonomy.

Don’t believe the hype

My social media and YouTube is packed full of content on how to “Automate your ENTIRE LIFE” with this AI Agent or “I made $200k with this agent and I’ll show you how”.

You know those LinkedIn posts promising that AI will automate your entire life by next Tuesday? Yeah, I’m tired of those too. The reality I’ve found is both more exciting and more nuanced than the hype suggests.

Here’s what I’ve learned from my experiments with AI agents:

First off, when they work, they’re absolutely mind-blowing. Imagine typing plain English instructions and watching an AI make decisions, update systems, and handle email responses - just like a smart team member would. No more dragging nodes around or setting up complex if-then scenarios. It feels like the future we were promised.

But here’s the thing - and I’m being completely honest here - it doesn’t always work smoothly. I’ve seen AI agents get stuck in infinite email loops (sending jokes to themselves!), struggle with simple date formatting, and sometimes just completely misunderstand what they’re supposed to do.

This mirrors what Greg Isenberg pointed out recently about the massive shift happening in SaaS. He argues that we’re witnessing the slow-motion collapse of traditional SaaS as AI agents start taking over. The $1.3 trillion industry is being hollowed out from within.

The transition looks something like this:

Phase 1 (Now): AI as co-pilot - making existing software more efficient Phase 2 (Next 12-18 months): AI agents becoming autonomous operators Phase 3 (2-3 years): Software becoming invisible as agents work directly with APIs

But here’s what I think is really interesting: while the big SaaS companies are bringing knives to what they think is a knife fight, we - the bootstrapped founders and indie hackers - have a unique opportunity.

Instead of building traditional SaaS products with rigid workflows and fancy UIs, we can focus on creating flexible, AI-first solutions that solve real problems. The barrier to entry is dropping dramatically - what used to cost £1M to build can now be done for a fraction of that with no-code tools and AI.

My advice? Start small and stay practical:

1. Don’t try to automate everything at once. Pick specific, non-critical tasks to experiment with AI agents.
2. Be prepared for some trial and error. Document what works and what breaks.
3. Focus on solving real problems rather than chasing the latest AI hype.

The most exciting part? This is just the beginning. Yes, there are limitations and frustrations right now. But the speed of improvement is incredible. What doesn’t work today might work flawlessly next month.

For us indie hackers and bootstrapped founders, this means we can build sophisticated solutions with a fraction of the resources traditionally required. We can compete with established players not by outspending them, but by being more nimble and creative with these new tools.