Use case · Chatbots and knowledge
We answer the same questions every day. The answers are already in our docs.
A chatbot built on your own documents answers customers or staff with your real policies, not guesses. Experts call this RAG: the bot first finds the right passage, then writes the answer from it. Willo matches you with experts who build and test these.
Sounds familiar? You are in good company.
Customers
“We get the same 20 questions over and over.”
New hires
“New people ask the same things for their first three months.”
Documents
“The answer is in our handbook somewhere. Nobody can find it.”
Help center
“We have a help center, but customers still email us.”
From a few lines to a clear brief.
This is the kind of thing a company types into Willo, and the brief that comes out of it.
What the company typed
“Our support team answers the same questions about delivery, returns and warranty every day. The answers are in our help center and a few PDFs. We want a chatbot on the website that uses those and hands over to a person when it isn't sure.”
Illustrative example — not a client case.
What a good expert does here. Good sources, tested answers.
Cleans up the source
The bot is only as good as the documents behind it. They check what is out of date, duplicated or missing.
Builds a test set
They collect real questions with the right answers. Every change is checked against that list before it goes live.
Designs the handover
When the bot is unsure, it says so and passes the chat to a person. It shows its sources so people can check.
Before the first call
What to have ready
- The documents it should answer from: help pages, policies, handbooks
- 20–30 real questions people ask, with good answers
- Where the bot would live: website, help desk or internal chat
- Who keeps the documents up to date
Be honest about
Things to watch
- Chatbots can sound sure and still be wrong. Test them and show sources.
- Keep private and internal documents out of a public bot.
- Decide what the bot must never answer, such as legal or medical questions.
Questions about chatbots, answered.
What is RAG?
RAG stands for retrieval-augmented generation. The chatbot first searches your documents for the right passage, then writes its answer from that passage. The answers come from your material, not from general internet knowledge.
Will the chatbot make things up?
It can, which is why testing matters. A good expert limits the bot to your documents, shows sources, and tests it against real questions before launch. When it is not sure, it should hand over to a person.
Can it be for our staff only?
Yes. An internal assistant can answer questions about your policies, processes and products. Access can be limited to your team.
Do our documents need to be perfect first?
No, but they need to be current. Part of the first step is finding gaps and outdated pages. Fixing those helps your team even before the bot is live.
Tell Willo which questions keep coming back.
A few sentences is enough. A real person reviews the brief, and you get 2–3 matched experts, usually within 3–7 days. The first step is a short call with no commitment. Free for companies.