The Carousel Became a Chatbot: How Client Demands Shifted to AI

A few years ago, the universal client request was a homepage carousel. That feature is now the AI chatbot. The pattern is the same: clients want the thing they saw on a competitor's site, regardless of whether it serves their users.

The post documents the experience of a freelance developer watching this demand shift happen in real time. Carousels were annoying because they were low-value and technically trivial — chatbots are a different problem. They carry genuine implementation complexity, ongoing API costs, and a maintenance surface that most small-business clients are not prepared for.

For solo developers, the scoping challenge is harder than the technical one. A carousel ships and sits. A chatbot backed by an LLM has latency, failure modes, prompt injection surface, and a cost model that scales with usage. Clients do not understand this distinction. They see a text box that answers questions and assume it behaves like a button.

The practical implications: project agreements now need to address API cost pass-through, acceptable-use boundaries, and what happens when the model hallucinates. None of this was in standard freelance contracts six months ago. Agencies writing new SOW templates are ahead; most solo freelancers are not.

There is also a positioning question. Developers who built carousels were interchangeable. Developers who can scope, build, and maintain a chatbot integration — including retrieval pipelines, context management, and guardrails — are not. The demand shift is real, but the skill gap between implementing a wrapper around a chat API and building something production-worthy is significant.

The author's framing maps cleanly to a broader observation: client demand has always moved faster than client understanding. What changed is the gap between what clients think they are asking for and what they actually need is now larger, and the cost of getting it wrong is higher.

Source