---
title: "Bot Knowledge, ProxyAI"
description: "Answers grounded in your documents"
canonical: "https://www.proxyai.app/services/rag"
language: "en"
---

> For the complete documentation index, see [llms.txt](https://www.proxyai.app/llms.txt). Request public content pages with `Accept: text/markdown`.

[All services](https://www.proxyai.app/services)

![Bot Knowledge](https://www.proxyai.app/_next/image?url=https%3A%2F%2Fpub-7b2322170fbe462d9923ad2fef945c5e.r2.dev%2Fasset%2F1785597873933-03336acd.jpg&w=3840&q=75)

# Bot Knowledge

 ADD-ON 

Answers grounded in your documents

 WHAT'S INCLUDED  Usage applies to credit balances  Hybrid keyword and schemetic search  Document editor 

Includes $19.00 of AI credits on purchase

 $38  one-time setup 

[Add to your stack](https://www.proxyai.app/services#builder)

## Works well with

[Abuse Guard Moderation & safety filters built in $10](<https://www.proxyai.app/services/abuse-guard>)

[Human Handoff A customer asks for a person, and one takes over — with a briefing already written $10](<https://www.proxyai.app/services/human-handoff>)

[Helpdesk Connect Your bot opens tickets in the helpdesk you already use — with the conversation attached $20](<https://www.proxyai.app/services/helpdesk>)

[Agent Tasks AI agents that run scheduled jobs on your store — reports in your inbox, knowledge base kept fresh $40](<https://www.proxyai.app/services/agent-tasks>)

[Sentry Visitor analytics on your site: which pages they read, where they click, and how many go on to chat $20](<https://www.proxyai.app/services/sentry>)

Retrieval

## Hybrid Semantic & Keyword Retrieval

Vector meaning + exact keyword search

Combines vector meaning with exact keyword search. Finds specific model numbers, variant SKUs, and exact phrasing without missing technical details.

OCR

## On-Device OCR for Scanned PDFs

Transcribed in your browser

Scanned paper guides, legacy spec sheets, and photo documents are transcribed directly in your browser. Clean text is indexed without sending images to third-party OCR APIs.

File formats

## Broad File Format Support (19 Formats)

Up to 10MB per file

Accepts PDF, DOCX, XLSX, CSV, TXT, Markdown, HTML, EPUB, and JSON up to 10MB per file.

Review

## Review Before It Becomes Searchable

Upload → Extract → Review → Implant

Extracted content is held for review before implantation, helping you keep outdated or irrelevant material out of the bot’s searchable knowledge.

FAQ

## Frequently asked questions

 How large can my documents be? 

Files can be up to 10MB each, containing up to 4MB of extracted text (~10 million characters). Oversized files can simply be split into smaller sections.

 How does billing work? 

 Do changes take effect immediately?
