Just published a guide on How to Create an AI Chatbot, and if there was one takeaway to share with someone who’s undecided on chatbots or who ‘tried, failed, and dropped’, it would be this:
The difference between a chatbot visitors ignore, and one they actually use comes down to the specifics — the right training content, clear instructions, and a setup that matches your use case.
What’s the one chatbot setup change you’ve made that had the biggest impact?(Could be training, could be a tiny UI change — interested in the non-obvious stuff.)
Deleted every instance of “totally understand,” “no worries,” and all emojis. It felt harsh at first, but completion rates went way up. People don’t trust a bot pretending to get them, they just want it to work and help. Switched to neutral, direct responses
Well everybody is different, and for me, I don’t think this would work out for my goals. I find that AI is good for disability access and certain tasks that traditional methods cannot fulfill, but I personally don’t like the idea of it replacing the parts that traditional methods can. But this is just my opinion. Thanks for sharing.
Not having a Chatbot on your website today is much like not having your employees use texting on their phones. Back then everybody thought a phone call was much better and/or leaving a voice message. …Get ahead of the times!
Thank you for sharing your perspective! It’s true that AI isn’t a one‑size‑fits‑all replacement for traditional methods, especially in areas where human empathy, judgment, or deep experience matters a lot. Curious – do you find there are specific scenarios where you would consider AI helpful alongside traditional methods?
Yes I do. So here are examples. My fiance is blind, and I learned through her how there is little access for the blind community on modern applications, since most of what we do on tech is visual. Thankfully, however, AI can help out. So in the Photos application for example, which is very visual, AI can describe what is going on in each photo. That way she can make notes on the side. But what can be difficult is the amount of photos means an extra long time to make notes, and there is currently no batch-AI tool to describe multiple photos at once. And usually what helps with blind access is when there is text based content, because then her screen reader can read the text. But images take extra work.
So if there was a batch AI describer, I would be all for that. I want to have more disability access on my website, and so that would work great for my fiance, if there was something that could describe the visuals of images and videos, and then generate notes automatically.
So in my opinion, I support AI tools to be used for things such as that, in which there is very little access, but I don’t support AI replacing any type of social interactions. But that is just my opinion. Anyone is free to disagree.
Thank you for sharing this – that’s a really thoughtful perspective.
What you described is exactly where AI feels most meaningful, helping fill gaps that traditional tools haven’t solved well, especially in visual content and accessibility. Sadly, much of the web still isn’t built with proper image descriptions.
At the same time, I’ve seen AI work well as a support layer, handling simple things quickly while humans step in where it matters. That balance seems to benefit both users and businesses when it’s done right (and for the right reasons). But to each their own, of course.
The difference between a chatbot visitors ignore, and one they actually use comes down to the specifics — the right training content, clear instructions, and a setup that matches your use case.
Poor friendly chatbot…No honestly, it gets annoying, doesn’t it? “Totally understand” - no you don’t, just help me