Chatbot
A chatbot is a program that holds a written conversation with a user. In support, current ones index the company documentation and answer from it; earlier ones followed decision trees written by hand.
What is chatbot?
A chatbot is a program that holds a written conversation with a user. In support, current ones index the company documentation and answer from it; earlier ones followed decision trees written by hand.
Also: virtual assistant · support bot
Three generations, one name
Three very different things get called a chatbot: rule-based ones, which follow a diagram and only understand what was anticipated; documentation ones, which search the company content and write with a language model; and agents, which also read data and take actions. When somebody says "we tried a chatbot and it did not work", the useful first question is which of the three.
Why it matters
What changes in a SaaS
It is the category everything else gets compared against, and the one that sets expectations. It is also where most of the disappointment happens, almost always for the same reason: a documentation chatbot deployed against a problem that needed the customer account, not an article.
Chatbot in detail
What it handles well
Questions whose answer is the same for everyone: how to do something, where an option lives, what an error message means. It takes an afternoon to set up and costs little, and for that it is the right choice.
Where the ceiling is
The moment the question depends on the account. "Why was I charged this?" is not written in any article, because the answer changes with who is asking. A documentation chatbot can only send the generic article, and the customer, who already read it, writes again.
The second message eats the saving
This is the effect nobody counts. If the bot answers and the customer rephrases, you have not automated a conversation: you have added a step to one that still needs a person.
How Intake handles it
Questions about chatbot
What is the difference between a chatbot and an AI agent?
The chatbot answers from what is written; the agent also reads the account state and can take actions. It is not about the quality of the model, it is about what data it can reach.
Is a chatbot enough for a SaaS?
For the share of questions whose answer is the same for everyone, yes. If most of your conversations start with "in my account…", no documentation chatbot will lower your volume however good it is.
Related terms
A term on its own is only half understood. These come up in the same conversation.
AI agent
An AI agent is a system that, beyond generating text, reads data and takes actions to finish a task. In support that means it checks the actual state of the customer account before answering, and when the answer involves doing something, it does it instead of explaining how.
RAG — Retrieval-augmented generation
Retrieval-augmented generation, or RAG, is the technique of finding relevant information in your own sources and handing it to a language model so it writes the answer with that. The model does not memorise the content: it looks it up at answer time.
LLM — Language model
A language model is a system trained on enormous amounts of text that, given a fragment, predicts how it continues. Writing, summarising, translating and holding a conversation all come out of that one simple capability: they are all ways of continuing a text plausibly.
Hallucination
A hallucination is an answer from a language model that sounds right and is not: an invented figure, a feature that does not exist, a plausible number nobody checked. It is not a bug, it is a direct consequence of how a model works — it produces the most probable continuation, not the verified one.
Deflection rate
Deflection rate is the percentage of questions resolved without a person getting involved, either because the customer found the answer themselves or because an automated agent closed it. It is the metric used to justify any investment in self-service or in AI.
Latency
Latency is how long an AI system takes to return an answer from the moment it receives the question. It includes the information retrieval, the lookups against other systems and the text generation, and it is perceived whole even though the model is only one part of it.
