There’s another thing worth looking at here. A good support AI shouldn’t just reduce the number of tickets, it should help you understand why customers are contacting you in the first place.

If the same question keeps coming up, maybe the problem isn’t the support team. Maybe the product is confusing or the documentation isn’t clear.

So after a few months, I’d look at more than just ticket reduction. I’d want to know what the AI is learning from those conversations and whether it’s helping us spot problems we didn’t even realize we had.

In that sense, the chatbot could become more than support ,it could be a window into what customers are struggling with.

One thing I’d look at that isn’t really mentioned is the “cost of babysitting the bot.” A $50 chatbot sounds cheap until someone has to constantly check its answers, fix its knowledge, handle weird conversations, and clean up mistakes.

So I wouldn’t only ask, “Which chatbot can I afford for $100?” I’d ask, “Which one can actually take work off my plate without becoming another task I have to manage?”

That difference probably becomes much clearer after a few months in the real world.

What interests me most is that Backstage could eventually make the whole idea of “managing software” feel outdated.Instead of hunting through settings to fix something, you could simply tell the agent what problem you’re seeing and let it figure out the steps, explain the reasoning, and ask for approval.

The real test, though, is when something goes wrong. If it can explain why it recommends a change instead of just making one, that’s where I think the real value is.

There’s another side to this that we don’t talk about enough. AI agents aren’t just helping companies automate work; they could actually become a way of preserving how a company thinks.People leave, teams change, and even good processes slowly get forgotten. An agent can keep that institutional knowledge alive and make it available whenever it’s needed.But there’s a funny catch, if you give an AI agent a bad process, it can become really good at doing the wrong thing consistently .

So the real value isn’t just automation or consistency. It’s forcing companies to finally ask, “Is the way we’re doing this actually the right way?” AI might end up exposing bad processes just as much as it improves them.

The bigger change might not just be the AI agent itself, but what we start paying for. If a SaaS product can actually get work done for me instead of just giving me tools to do the work myself, I’ll probably care less about how many features it has and more about how much time and effort it saves me.That could really change how SaaS companies compete. Instead of saying, “Look at all the features we have,” they may eventually be saying, “Here’s how much work we can take off your hands.” And honestly, I think that’s a much more interesting direction for SaaS.

The interesting thing about a waitlist is that it can measure excitement, but not necessarily commitment. Someone can join a list because the product sounds interesting and then completely forget about it a week later.

What I’d be more curious about is whether you can create a “soft waitlist” after launching. Let people use the product immediately, but give early users a reason to stay involved early access to features, a small community, or even direct conversations with the team. That way you get the speed of launching publicly while still building the community and anticipation a traditional waitlist was supposed to create.

The part teams might overlook is that AI customization can become a kind of “invisible debt.” Everything works fine at first, so you keep adding prompts, rules, and integrations without thinking much about the future. Then a model update happens and suddenly something that worked perfectly starts behaving differently.

The goal shouldn’t be to make the AI as customized as possible, but to make it easy to change when things change. I’d keep the custom pieces that users actually notice and simplify everything else. A little less customization today might save the team a lot of headaches six months down the road.

I don’t think skipping the waitlist was necessarily a mistake. Sometimes a waitlist can create the illusion that a product has demand simply because hundreds of people dropped their email. The more interesting test is what happens after you give them the product,do they use it, come back, pay for it, or tell someone else about it.

I’d probably launch early if the product is usable, and I’d treat the first users almost like a research team. Their questions, complaints, and even the features they totally ignore can tell you more than a long list of people waiting. You can always build hype later, but you can’t replace the lessons that come from watching real people use your product.

From the above discussion the real question is not just whether AI converts better than live chat, but what makes people trust the interaction enough to actually buy. An AI agent can make things easier by understanding what a visitor needs and guiding them to the right option instead of just answering questions. But if the responses feel too robotic or inaccurate, people may quickly lose confidence. So I feel the quality of the conversation and knowing when to involve a real person may matter more than simply having an AI agent on the website.

The bigger change is AI becoming the middleman between us and businesses. Instead of going through hundreds of products ourselves, we could simply tell AI what we want and let it narrow things down for us. That means that businesses will need to make their product information clear, accurate, and trustworthy. But it also makes me wonder what happens to good products that AI never discovers.