A new essay by a web designer argues that AI chatbots have become the carousel of 2026: a feature businesses request because competitors have one, then never measure. The piece, which trended on Hacker News this week with more than 165 points, describes a familiar pattern. A client wants a chatbot. They cannot say what it should do. One competitor’s chatbot reportedly gave out wrong opening hours for months without anyone catching it.
The observation lands in Australia at an awkward moment, with SMEs under real pressure to look ‘AI-ready’ but few with the analytics culture to validate the spend.
What the essay actually says
The essay, written by designer Adële, walks through three eras of must-have website features: image carousels in the early 2010s, cookie banners after GDPR, and now AI chatbots. Her complaint is not about the technology. It is about the buying behaviour. Clients ask for chatbots because they have seen one elsewhere. They rarely define what the bot should do, who should use it, or how success will be measured.
The author noted one client who had not opened the analytics dashboard in 18 months and could not remember the login. The trend, in other words, is feature mimicry dressed up as digital strategy.
Why the comparison sticks
Carousels failed in measurable ways. Eye-tracking studies in the 2010s repeatedly showed visitors tuning them out. They survived because they made homepages look busy. AI chatbots are tracking the same arc: high deployment, low scrutiny, easy to install, hard to make useful.
A bot that hallucinates opening hours or loops a customer back to a generic greeting is not a small bug to be fixed later. It is a customer service incident that runs around the clock until somebody notices. The cost of an unwatched carousel was a slow homepage. The cost of an unwatched chatbot is a damaged reputation.

Why Australian businesses should pay attention
For Australian SMEs, the pressure to add AI features keeps building. Competitors are doing it. Industry events keep selling the urgency. Yet the OAIC and consumer law principles around accurate information apply just as much to a chatbot’s output as to a print advert. A bot that misrepresents pricing, returns policy or compliance information can create the same liability.
The right starting question is not ‘should we add a chatbot’ but ‘what specific customer task would this solve, and how will we know if it works’. For some businesses with high inquiry volume, predictable questions or after-hours demand, the answer points cleanly to AI chatbot development. For others, a sharper FAQ page does the same job for a fraction of the cost.
What measurement should look like
The essay’s hardest finding is not technical. It is that clients buy and then forget. A chatbot worth its budget needs three things in place from day one: a defined success metric like deflection rate or conversation completion, a weekly check on transcripts to catch hallucinations early, and an owner inside the business who is accountable when something breaks.
Australian companies that have done this work, including Canva’s recent AI Discovery Week, started with internal capability before customer-facing automation. The lesson holds at any scale. Map the spend against alternatives first. Sometimes broader AI consulting work identifies where automation actually moves the numbers, not just the homepage.
A more useful first question
The carousel era ended not because clients learned to read analytics but because designers stopped offering them. That correction is unlikely with chatbots. They are too embedded in the AI procurement story.
What can change is the order of the conversation. Before the bot, define the customer task. Before the budget, define the metric. Done well, an AI chatbot is a serious operational asset. Done as decoration, it is a 2026 carousel that argues with your customers.
