One thing I see a lot is AI suggestions being added everywhere just because they can be. Sometimes I’m doing a simple task and the AI adds another step instead of removing one. If the feature doesn’t save time or make the workflow easier, I’d rather have it stay out of the way.

The part about AI agents actually completing tasks is what interests me most. A chatbot can answer a question, but being able to check an account, update information or solve an issue without passing the customer between different systems feels much more useful. I think that difference will become clearer once more companies start using these tools in real customer support environments.

I’ve noticed the same thing with smaller SaaS products. Some of them solve a very specific problem really well but don’t get much attention because they aren’t chasing the usual startup trends. I usually find these tools through recommendations or by coming across them while looking for a solution to a particular problem. Those are often the products that stick with me.

I’ve seen this happen where the AI feature sounds great during planning, but once it’s live, users don’t really use it as much as expected. The maintenance and constant tweaking can also become a bigger part of the workload than people expect. I’d probably start small, see if customers actually rely on it, and only then invest heavily in building it out.

I can see AI agents becoming a bigger part of online shopping, especially when it comes to comparing products and finding reliable information. Instead of going through several websites, people could just explain what they need and let the AI narrow down the options. SEO would still matter, but the way brands make their products discoverable could change quite a bit.

.

• This reply was modified 1 month, 2 weeks ago by  Ashyra.

What interests me most here is the difference between getting a deployment running and actually maintaining it over time. A tool can feel very convenient at the beginning, but production use brings a completely different set of concerns around updates, backups, monitoring, security, and troubleshooting. I’d also be interested in hearing how much hands-on maintenance people ended up doing after their initial setup.

I’d probably spend more time understanding the target customers before building too much. Having a clear idea of who the product is really for could make the messaging, pricing, and even the features much more focused from the beginning.

I think AI agents are useful when they actually remove extra work rather than just being added because they’re trending. For SaaS products, reliability and ease of use still matter most, and AI should support those things instead of making the product more complicated.

AI is definitely making online shopping more convenient, especially when it comes to finding relevant products and comparing different options. I also like that it can help businesses understand customer needs better while making the overall shopping experience simpler and faster.

I agree that the biggest challenge is often making AI feel like a natural part of the existing product. The technology can work well, but if it adds extra steps or doesn’t solve a real customer problem, people may simply continue using the product the way they always have.

I think the biggest benefit is reducing the friction between a customer having a question and actually making a purchase. If Chatbase can give accurate product recommendations and quick answers, it can make the whole process feel much smoother. But for returns, complaints, or unusual situations, having a human step in is still important.

I agree, especially with the point about AI not being able to fully understand human behavior. AI can spot patterns and qualify leads, but there’s still a big difference between recognizing a pattern and understanding why someone actually wants to buy. The best approach is probably to let AI handle the repetitive parts while keeping humans involved when the conversation becomes more complex or personal.

I think the real test starts after launch. If people are calling it “vibe-coded”, I’d take that feedback seriously rather than getting defensive. Find out what users actually dislike, fix the biggest pain point, and keep improving from there. A simple product that solves a real problem is always better than a polished product that nobody needs.

I’d say the biggest difference would be an AI agent that can actually understand what I’m looking for and narrow down the options for me. Just recommending products isn’t enough if I still have to compare everything myself.