AI Chatbot

The parts that never made it into the process document.

None of this is a technology problem, which is why buying technology has not fixed it.

1. 01

   It is confident and wrong, and there is no review of which answers were wrong, so the error rate is unknown rather than low.

2. 02

   There is no escalation path. A customer who is stuck has one option, which is to leave.

3. 03

   What it told a customer is not recorded against that customer, so the promise it made is discovered when the complaint arrives.

4. 04

   It is measured on deflection rate, which pays it to avoid helping.

Narrow, boring, and in your environment.

Scoped against a boundary agreed in writing before anything is built, and built in increments you can evaluate one at a time.

• Answers grounded in your own material, with the source shown, and a refusal when there is no source.
• Escalation to a named person with the conversation and the context already attached.
• A log of what was said to whom, and a weekly sample reviewed by a person, because that is the only way an error rate is ever known.

What the agent is not allowed to do here.

Every piece of this work has a line the automation does not cross, and it is cheaper to agree it now than to discover it in an incident. These are ours in ai chatbot.

• It does not guess. No source, no answer, and it says so.
• It never commits you to a price, a date or a refund. It asks.
• It is measured on resolved and correct, not on contained.

The six that apply to every engagement, whatever it is, are on the home page, and what an engagement costs is on pricing.