You probably don’t need a chatbot

The most common first request we hear is "we want a chatbot on our website". It is a reasonable instinct — it is the most visible form of AI. But when we map how the business actually works, the website chat is rarely where the value is.

Start from the pain, not the tool

A services company came to us for a chatbot. After a week of mapping their operations, the real problem was elsewhere: their two-person support team spent most of the day answering the same twenty questions by email, days late. The fix was not a widget on the homepage — it was an AI layer inside their ticketing that resolved the recurring cases and drafted answers for the rest.

Same technology, completely different placement — and measurable value: more than half of the tickets handled automatically, first response in seconds instead of days.

Three questions to ask before any AI project

• →Which task eats hours every week and follows a pattern?
• →Which number would we want that task to move — revenue, delay, error rate?
• →Who has to adopt the tool for the value to exist, and what do they need?

If a project cannot answer these three questions, it is a technology looking for a problem. If it can, the tool choice — chatbot or not — becomes obvious, and usually boring. Boring is good. Boring gets adopted.