Now with persistent pipelines & auth

Build RAG pipelines
in minutes

Upload your documents. Configure chunking, embeddings, and retrieval. Chat with your data — powered by Google Gemini. No infrastructure to manage.

How It Works

Four steps to your own RAG pipeline

01UPLOADPDF, TXT, DOCX
02CHUNKSplit & embed
03INDEXVector store
04CHATAsk anything
RAGForge
RAGForge
Pipeline Active
+ New Pipeline
What were the key findings from the Q4 report?
Based on the Q4 report, the key findings include: 1. Revenue grew 23% year-over-year, exceeding the projected target of 18% 2. Customer retention improved to 94.2%, up from 89.7% in Q3 3. The new onboarding flow reduced time-to-value by 40%
q4-report.pdfmetrics-summary.docx
How does that compare to Q3?
Comparing Q4 to Q3: • Revenue: Q4 saw 23% growth vs Q3's 15% — an 8-point acceleration driven primarily by enterprise tier upgrades • Retention: 94.2% vs 89.7% — the 4.5-point jump correlates with the support restructuring completed mid-Q3
q4-report.pdfq3-report.pdf
>Ask a question...
Features

Everything you need, nothing you don't

Multi-Format Ingest

3

file formats

Upload PDFs, DOCX, and plain text. Documents are parsed, chunked with configurable overlap, and embedded using Gemini's latest embedding model.

Persistent Pipelines

sessions

Your pipelines are saved to Supabase. Come back anytime, load a pipeline, and continue chatting where you left off. Full history preserved.

Vector Search

768

dimensions

Embeddings stored in Supabase pgvector with IVFFlat indexing. Cosine similarity retrieval finds the most relevant chunks for every query.

Streaming Responses

<1s

first token

Gemini responses stream in real-time. See answers appear token-by-token with source attribution showing exactly which documents were referenced.

Configurable Pipeline

6

parameters

Tune chunk size, overlap, top-K retrieval, model selection, and system prompts. Every parameter is adjustable to match your use case.

Secure by Default

0

keys stored

Clerk authentication protects every route. Your Gemini API key is used per-request and never stored. Each pipeline is scoped to your account.

Why RAGForge

Skip the boilerplate

 DIY SetupRAGForge
Vector DB setupHoursBuilt-in
Document parsingWrite your ownDrag & drop
Embedding pipelineCustom codeAutomatic
Auth & multi-userIntegrate yourselfClerk included
Chat with sourcesBuild from scratchReady to use
Time to first queryDaysMinutes

Ready to forge?

Create your account, bring your Gemini API key, and start chatting with your documents in minutes.

Get Started — Free