Skip to content
Train on Your Data

Train an AI Chatbot on Your Documents and Website

Upload PDFs and other documents, paste page URLs or add your whole website. Botcadence indexes your content, understands context and nuance, and answers from your actual documentation instead of generic responses.

Updated

Playground

Test before launch

PDFs

Docs & Sites

Grounded

Responses

Auto

Content Refresh

In short

What is a Document Training?

Document training lets you upload PDF, DOCX, TXT, CSV and Markdown files, add pages by URL or crawl your whole website, so Botcadence answers from your approved knowledge, not generic AI. Websites are recrawled automatically, and the starter templates tell the agent not to guess when nothing covers a question.

The Problem

Why Knowledge Never Reaches Customers

Train Botcadence on your approved documents so answers stay grounded, current, and ready across channels.

Knowledge Lives in Scattered Docs

Help centers, PDFs, and Notion pages never make it into agent workflows consistently.

Scripted Bots Need Constant Rewrites

Intent trees break whenever pricing, policies, or product details change.

Ungrounded Answers Create Risk

Generic LLMs invent policies instead of citing approved business content.

Training Takes Engineering Time

Teams wait on custom builds instead of uploading the sources they already trust.

Key Benefits

Why Document Training Matters

Multi-Format Ingestion

Upload PDF, DOCX, TXT, CSV and Markdown files, add single pages by URL, or add your whole website to crawl. Export Google Docs, Notion or Confluence pages to add them.

Files, pages and whole sites

Search by Meaning

Search combines meaning and keywords, so the agent finds the right passage even when a customer words a question differently from your docs.

Meaning plus keywords

Grounded Responses

Answers come from your source material, and the starter templates tell the agent to say so instead of guessing when nothing covers a question. Add a handoff step to pass those questions to your team.

Answers from your sources

Website Recrawls

Add your website and Botcadence recrawls it automatically, every 30 days by default, so answers follow your published pages. Replace an uploaded file when it changes.

Scheduled recrawls

Source List

See every file, page and website the agent learns from, with its sync status, and delete or replace any of them when your content changes.

You control the sources

Gaps From Real Conversations

Read conversations in the inbox to see where the agent handed off or said it didn't know, then add the document that answers it.

Spot knowledge gaps

How It Works

How Document Training Works

Step 01

Upload Your Content

Upload PDF, DOCX, TXT, CSV or Markdown files, paste page URLs or add your whole website. Export pages from Notion, Confluence or Google Docs to add them.

Step 02

AI Processes & Indexes

Botcadence splits your content into passages and indexes them for search, so the agent can find the passage that answers each question.

Step 03

Test & Validate

Ask questions in the built-in playground before going live. Spot gaps and fix them by adding documents or editing your instructions.

Step 04

Deploy & Monitor

Go live, read conversations in the inbox and add documents wherever an answer falls short.

See Your AI Trained on Your Own Docs

Upload a sample document during your demo and watch the AI answer questions from it in real time.

Example workflow

New return policy question answered accurately on day one

You publish an updated warranty document. Botcadence indexes the file, and the next customer question about extended coverage gets an answer from the new policy.

1

Trigger

Customer asks about extended warranty eligibility

A buyer chats in asking whether their purchase from six months ago qualifies for the newly announced two-year extended warranty program.

2

AI action

Agent retrieves answer from freshly indexed policy doc

Botcadence finds the warranty passage in your uploaded PDF and states the eligibility window, exclusions and claim steps from it.

3

System update

Answer saved in the transcript

The full conversation is saved in the inbox, so compliance can read exactly what the agent told each customer.

4

Business outcome

Consistent policy answers reduce disputes and escalations

Customers receive accurate information aligned with legal-approved docs, cutting incorrect promises that create costly make-goods later.

The Transformation

Scripted Bots vs Document-Trained AI

See how grounding agents in your documents beats brittle decision trees and generic LLM answers.

CapabilityWithout AIWith Botcadence
Setup EffortMonths of intent scriptingUpload docs and test answers in the playground
Answer GroundingGeneric model guessesResponses tied to approved sources
Content UpdatesRebuild flows for every changeRefresh training from source docs
CoverageOnly scripted FAQsLong-tail questions from your corpus
AuditabilityHard to prove what was saidFull transcript of every conversation

Common Questions

Document Training Frequently Asked Questions

Botcadence supports PDF, DOCX, TXT, CSV and Markdown files, single web pages by URL and whole websites, which are recrawled automatically every 30 days by default. There is no live connector to Notion, Confluence, Google Drive or SharePoint, so export pages from those tools as PDF or DOCX, or add their public URLs.

Indexing starts as soon as you add a source, and each file, page or website shows its status in your source list. Larger websites take longer because every page is crawled and indexed. How much you can add depends on your plan's page and upload limits.

Botcadence searches your documents for each question and gives the AI the matching passages to answer from. The starter templates tell the agent not to guess, and you control what it does when your sources do not cover a question, such as saying so or handing off to your team. Test in the playground and read conversations in the inbox.

Yes. Upload a new document or replace an old one and only that source is indexed again, usually within minutes. Websites you add are recrawled automatically, every 30 days by default.

Data is encrypted with AES-256 at rest and TLS 1.2+ in transit. Your documents are only searched for your own projects, and Botcadence does not train or fine-tune AI models on them. Ask us for details during your security review.

Get Started Today

Give Your AI the Knowledge It Needs to Deliver

Upload your documents, add your website and deploy an AI that answers from them, without months of scripting.

  • Free plan, no credit card
  • Website chat live in about 5 minutes
  • Works with your existing tools

Ask AI about this page