kilo-logo
kilo-logo
kilo-logo

Knowledge base chatbot

Knowledge base chatbot

Knowledge base chatbot
that cites its sources.

Knowledge base chatbot that cites its sources.

Fetch Hive is an AI workflow and agent builder. Use it to turn your website, documents and notes into a knowledge base, then give it to an agent that answers questions with quoted passages and a link to the source page, so people get the answer that is in your docs and not a guess.

Free Developer plan · No card required · Ask the live chatbot on this page without signing up

Why build one

Why build one

Most questions are already answered somewhere in your docs

Customers, new hires and teammates ask things your docs already cover, and somebody still stops to look it up and paste the link. A knowledge base chatbot searches your content for the passages that match, answers from them and shows where each answer came from. When the docs do not cover it, it says so.

Support teams

Let the chatbot take the questions your help centre already answers, in seconds and with the source page attached, so the queue is left with the ones that need a person.

Docs and knowledge owners

Every question and answer is logged, so you can see what people ask most and where the docs came up empty, and know which page to write next.

Small teams without a support desk

Share the chatbot with a link and questions get answered from your docs at 3am, with the source shown, while nobody on the team is online.

RAG chatbot

RAG chatbot

Four jobs behind a RAG chatbot that answers from your content

A RAG chatbot is only as good as the knowledge base behind it. These are the four parts you set up in Fetch Hive with no code: the agent that answers, the content it reads, the search that finds the right passage and the check on what is missing.

Answer questions with quoted sources

Give an agent a knowledge base and it runs a Vector Search for each question, answers only from the passages it finds and links each one back to its source page. If nothing matches, it says the docs do not cover it.

Build a knowledge base from a website, files and text

Add sources three ways: scrape a website, upload documents or paste in custom text. Fetch Hive splits them into chunks and indexes them, so one knowledge base can hold your help centre, your product PDFs and your internal notes.

Test and tune the search

Run a test query and see which passages come back and how they score. Choose Hybrid, Vector or Full Text search, set how many chunks to return and raise the score threshold until the weak matches drop out.

Find what your docs do not cover

A workflow answers a customer question from a public help centre page with a source for every claim, and lists the parts of the question your docs do not cover, so you know which article to write next.

Try it live

Try it live

Chat with a real knowledge base chatbot right here

This is the RAG Chatbot agent template running on this page. It is connected to a knowledge base built from the Fetch Hive docs, 308 pages in all. Ask it anything about Fetch Hive and it answers with quoted passages and links to the source pages. Three questions are free with no signup.

Chat with the Fetch Hive docs

A Fetch Hive agent with one tool, Vector Search, and one knowledge base, our own docs. Ask it anything about Fetch Hive. Every answer shows the passages it came from and the pages they sit on.

Knowledge base: Fetch Hive docs308 pages indexedVector Search
Example chat

An example chat from a real run. Ask your own question below to start a live one.

How many tasks does a knowledge base search cost, and can Vector Search run on a schedule every morning?

Docs agent2 passages from 2 pages

A Search Knowledge Base call costs 1 fixed platform task per request. The docs I found do not say that Vector Search can be scheduled to run every morning, so I can't confirm that from the docs.

Passages the answer came from

“Search Knowledge Base | 1 | Fixed platform task credit cost.”

1Resources: Task costsQuote found in the knowledge base

“- Build support or ops agents that search a curated Knowledge Base. - Retrieve source material before the agent writes a response.”

2Agents: Vector Search · Use casesQuote found in the knowledge base
Not in the docs. The docs found here do not say whether Vector Search can be scheduled to run every morning.

Source pages

16.4sEvery search and token is logged when you run this agent in Fetch Hive.

Try a question

Your first 3 questions are free. No account, nothing installed. It answers from the docs only.

Want this chatbot on your own docs?

This is a Fetch Hive agent, not a black box. Clone it, attach a knowledge base built from your own docs, files or site, and put it behind your help centre, your app or Slack.

Prefer chat? Run it from Claude or ChatGPT, free as a guest for 72 hours.

Clone & Automate

How it works

How it works

How this agent answers, step by step

The chat above is one Fetch Hive agent with a knowledge base attached. Every agent template works the same way, so you can clone it, swap in your own knowledge base and change the instructions.

See what every step does and what it costs

Every run of this workflow in Fetch Hive logs all of this, so you know what happened and what it cost.

  • Every prompt
  • Every model call
  • Every tool call
  • Every workflow step
  • Every input and output
  • Every cost
Example run log7 of 7 steps logged
A typical run of this workflow costs about $0.01.Cost so far $0.01

Clone it, then swap the knowledge base

Attach your own knowledge base, rewrite the instructions or change the model. It is your copy, so nothing is locked.

Share it with a link or Live Chat

Open the agent in Live Chat or publish a public share link, so teammates or customers can ask it questions without you building a front end.

Call it from your own app

Every agent can be called through the Fetch Hive API and SDK, so the same cited answers can appear inside your product or your help centre.

Free AI tools

More AI tools you can try free

Each tool below is a Fetch Hive workflow or agent running on this site. Try it with no signup, then clone the template behind it.

Knowledge base templates

Start from a template that works from your docs

Every template opens as your own copy. Change a step, swap a model, connect your apps and run it.

Workflow or agent

Workflow or agent

Should your knowledge base chatbot be a workflow or an agent?

Both run on the same models, tools and connected apps. Pick by how predictable the job is.

Workflow

The same steps on every run

Best for fixed jobs, like drafting an answer for every new ticket or adding new pages to the knowledge base

Uses Search Knowledge Base and Write Knowledge Base steps you can open

Runs on a schedule, from a webhook or through the API

Knowledge base agent

Works out the steps itself

Best for open conversations, where the next question depends on the last answer

Decides when to run a Vector Search and what to search for

Give it instructions and a knowledge base, with no code

You can mix them: add an agent as a step inside a workflow when one part of the job needs judgement.

Automate it

Automate it

Turn this demo into a chatbot on your own docs

Clone the agent behind the chat above, attach a knowledge base built from your content and share it. You will see every search it runs and what each answer costs.

Nightly docs sync

Scheduled

Every night

Scrape the docs pages that changed

Write

Add them to the knowledge base

Notify

Post what was added to Slack

Use it in your AI assistant

Run the RAG chatbot right inside Claude

Connect Fetch Hive once, then ask for this report in the chat you already have open. Free for 72 hours as a guest, no account needed.

Which AI assistant do you use?

Connect ClaudeGuest access, free for 72 hours
  1. Open Connectors

    In Claude (web or Desktop), open Settings, then Connectors, and click Add custom connector.

  2. Paste the connector URL

    Name it Fetch Hive and paste this URL.

    https://connect.fetchhive.com/mcp
  3. Connect as a guest

    Click Connect. When Fetch Hive opens in your browser, choose Continue as guest (or sign in), then click Allow.

Then ask Claude

Search my docs knowledge base and answer this with quoted sources: How do I rotate an API key?

Create a free account any time to keep what your assistant built. Stuck? Read the setup guide.

FAQ

FAQ

Frequently asked questions

Common questions about knowledge base chatbots, RAG and what it costs to run.

A knowledge base chatbot is a chatbot that answers from your own content and not from what a model happens to remember. It searches a knowledge base built from your docs, files and notes, answers from the passages it finds and shows the source. In Fetch Hive you build one as an agent with a knowledge base attached.

Still have questions?

Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image

Stop answering what your docs already cover.

Start from an agent that already works.

Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image

Stop answering what your docs already cover.

Start from an agent that already works.

Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image
Agent Image

Stop answering what your docs already cover.

Start from an agent that already works.