What is chatbot?

Why it matters

The word chatbot carries a decade of bad customer experience. Scripted bots trained an entire generation of buyers to type agent immediately. Understanding the difference between a decision tree and a grounded AI agent matters commercially, because they fail in completely different ways and are sold at similar prices.

What to know

• Rule-based bots are deterministic and auditable, which is genuinely valuable for regulated or high-risk flows.
• They break the moment a customer phrases something outside the script, which is most of the time.
• Maintenance cost is the hidden expense: every product change means rewriting flows by hand.
• Menu bots are effectively navigation, not support, and should be measured as such.
• A bot that cannot escalate cleanly does more damage than no bot at all.

An example

Example: a scripted bot offers four buttons. A customer whose issue is none of the four types a paragraph explaining their problem. The bot repeats the four buttons. The customer leaves and the interaction is recorded as deflected.

Common mistakes

• Deploying a scripted bot on open-ended question types
• Hiding the escalation path to protect deflection numbers
• Underestimating the ongoing cost of maintaining decision trees
• Assuming customers will rephrase to fit your script

Questions