Your Documents. Your AI.
Your Privacy.
Upload your company documents and ask questions. Kerdos AI answers strictly from your data — powered by LLaMA 3.1, FAISS vector search, and enterprise-grade RAG. Your data never leaves your environment.
I'd go with: Problem → Solution → Timeline → Ask. Keep it tight — one page max. The trick is making the problem feel urgent before you pitch the fix.
I found a few strong examples. The best ones all lead with a sharp metric — like “We're losing 12 hours/week to manual reporting.” Then they tie it to a dollar figure before pitching the fix.
Try It Right Now
Use the embedded Gradio UI or walk through the REST API yourself ΓÇö step by step.
Gradio Demo
Demo requires a Hugging Face API token with write access. For a fully private deployment, contact us.
API Playground
Walk through each API step live ΓÇö create a session, upload a file, and ask a question. All calls hit the real REST API at kerdosdotio-kerdos-llm-rag-api.hf.space
Each session gets an isolated FAISS vector index. Sessions auto-expire after 60 minutes.
Supported: PDF, DOCX, TXT, MD, CSV (max 50 MB)
Get a free token at huggingface.co/settings/tokens
Conversation will appear here after you ask a question.
API Log
Everything You Need
Built for enterprise document intelligence ΓÇö not just a chatbot.
Upload PDF, DOCX, TXT, MD, and CSV files ΓÇö all parsed and indexed automatically.
Answers are generated only from your uploaded documents ΓÇö no hallucination from internet knowledge.
Upload and query across multiple files simultaneously with a unified index.
Maintains full conversation context across questions ΓÇö natural dialogue, not one-shot Q&A.
CPU-friendly embeddings with all-MiniLM-L6-v2 + FAISS. Runs without expensive GPU hardware.
Files are processed in-memory and never stored after your session ends.
RAG Pipeline Architecture
A retrieval-augmented generation pipeline that grounds every answer in your documents.
Upload documents → parsed & chunked into 512-char segments with 64-char overlap
Chunks are embedded using all-MiniLM-L6-v2 and stored in a FAISS in-memory index
Your question is embedded → Top-K most relevant chunks fetched via cosine similarity
LLaMA 3.1 8B receives only the retrieved chunks and generates a grounded, cited answer
One API. Every Industry. Any Document.
The Kerdos AI RAG API integrates into your existing workflows in under an hour. Upload your proprietary documents and get hallucination-free, grounded answers — privately, at scale.
Healthcare & Pharmaceuticals
Instant answers from clinical trial documents and drug monographs
Challenge: Clinical teams spend hours searching regulatory submissions, pharmacovigilance reports, and RCT data.
Solution: Index clinical trial documents and drug monographs once. Query them in milliseconds with natural language — strictly grounded in your own data.
What to upload
POST /sessions/{id}/chat
“What are the contraindications for Drug X in diabetic patients?”
Legal & Compliance
AI-powered contract review with zero data leakage
Challenge: Reviewing hundreds of pages of contracts, NDAs, and regulatory filings is slow and error-prone.
Solution: Upload contracts and compliance documents. Get instant, citation-grounded answers on obligations, risk clauses, and deadlines. Your data never leaves your environment.
What to upload
POST /sessions/{id}/chat
“Summarise the indemnity clause in Schedule 3”
Banking, Financial Services & Insurance
Query RBI circulars, Basel III frameworks, and internal audit reports
Challenge: Internal teams need quick access to policy documents, investment frameworks, and regulatory circulars — accurately.
Solution: A private, grounded LLM that answers only from your internal documents. No hallucinations. No external API leakage. Full audit trail for compliance.
What to upload
POST /sessions/{id}/chat
“What is the capital adequacy ratio threshold per this policy?”
Manufacturing, EPC & Infrastructure
Field engineers query SOPs and safety manuals in natural language
Challenge: Field engineers need operational manuals, safety SOPs, and equipment specs on demand — often on-site without reliable internet.
Solution: Deploy the RAG API on-premise. Engineers query manuals using plain English — even in air-gapped environments — and get step-by-step, grounded answers.
What to upload
POST /sessions/{id}/chat
“What is the emergency shutdown procedure for Boiler Unit 4?”
Government & Public Sector
Instantly search tender specifications and statutory documents
Challenge: Government departments manage large volumes of statutory documents, tender specifications, and RTI responses that are slow to navigate.
Solution: Index tender documents and policy circulars. Enable officers to query eligibility conditions, submission procedures, and compliance requirements in seconds.
What to upload
POST /sessions/{id}/chat
“What is the EMD amount and submission deadline for this tender?”
HR, L&D & Organisational Knowledge
Give every employee a self-service AI assistant for HR queries
Challenge: Employees repeatedly ask the same questions on leave policies, onboarding procedures, and compliance norms — wasting HR bandwidth.
Solution: Index your employee handbook, POSH policy, payroll SOPs, and HR circulars. Deploy a private internal chatbot that answers only from approved HR documents.
What to upload
POST /sessions/{id}/chat
“How many earned leaves can I carry forward this year?”
Interested in a tailored enterprise pilot? partnership@kerdos.in
Open-Source, Battle-Tested
Embed Kerdos AI on Any Website
A lightweight JavaScript widget backed by the Kerdos AI RAG API. Add a floating document Q&A chatbot to your product, knowledge base, or internal portal with a single <script> tag.
<!-- Kerdos AI Widget --> <script src="https://cdn.kerdos.in/ai-widget.js" data-session="auto" data-theme="dark" data-primary="#0ea5e9" > </script>
One tag. Instant document Q&A.
Full Power. Full Privacy. Full Control.
The demo gives you a taste. The enterprise edition gives your organisation complete data sovereignty.
Private LLM Hosting
On-premise or private-cloud deployments ΓÇö your data never reaches any external API.
Custom Model Fine-tuning
Models fine-tuned on your domain data for dramatically higher accuracy on your content.
Data Privacy Guarantees
Complete isolation. Zero external data transfer. Full audit logs.
White-label Deployments
Fully branded for your organisation ΓÇö your name, your logo, your product.
Help Us Build the Enterprise Edition
We're raising investment to build the fully customisable enterprise edition ΓÇö private on-premise LLM deployments, custom fine-tuning, and white-label solutions for Indian and global enterprises.
Kerdos Infrasoft Pvt. Ltd. (CIN: U62099KA2023PTC182869) ┬╖ Bengaluru, Karnataka ┬╖ Est. December 2023