ToolsWallet RAG Platform Overview
Standard AI models often guess or make up facts (hallucinate) when asked questions about your specific business data. ToolsWallet RAG solves this by connecting your private documents (JSON datasets, FAQs, policy notes, and guides) directly to high-speed AI inference. It searches your indexed records, pulls the exact matching facts, and generates precise, cited answers with zero guesswork.
How to Use ToolsWallet RAG
From raw documents to a live website AI chatbot in 3 minutes.
Add Knowledge
Upload a JSON file or paste FAQ text (up to 25 Records). Checked with SHA-256.
Get API Key
Generate tw_rag_live_... and set CORS domain whitelist.
Deploy & Stream
Paste 1-line <script> on your site or call streaming REST API.
What is ToolsWallet RAG?
Retrieval-Augmented Generation (RAG) is a technique that gives AI models access to your private company data before answering questions. ToolsWallet RAG provides a ready-to-use, zero-cost knowledge engine with instant ingestion and real-time streaming answers.
- •Instant Document Ingestion: Files and JSON data are parsed, deduplicated, and available for search in under 50 milliseconds.
- •Complete Grounding & Citations: Every answer shows the exact document names and text passages used to build the answer.
- •Zero Hallucinations: The AI assistant strictly confines its answers to your verified knowledge base.
How to Use the System (4-Step Flow)
Any user or developer can set up an AI assistant in 4 simple steps:
- •Step 1 (Create Knowledge Base): Go to /dashboard/create and enter a bot name (e.g. 'Customer Support Bot').
- •Step 2 (Add Knowledge): Upload a .json/.txt/.md file or paste text snippets (up to 25 records per RAG).
- •Step 3 (Get API Key): Go to API Keys tab to generate a secure live key and whitelist your website domain.
- •Step 4 (Deploy): Paste the 1-line <script> tag on your website or call POST /api/v1/query/stream from your code.
Understanding Sources vs Records
To keep things simple and clear, our system uses two basic terms:
- •Source (File / Document): An uploaded file or text snippet (e.g. faq.json, return_policy.txt, or pricing.md).
- •Records (Extracted Items): The individual Q&As, products, or knowledge passages parsed from that source (e.g. 1 faq.json file might contain 4 Q&A records).
Each Free Tier Knowledge Base allows up to 25 Records. If you upload a file with 10 FAQs, it uses 10 of your 25 available record slots.
Engine Performance & Capabilities
ToolsWallet RAG is built for high reliability, ultra-low latency, and exact factual accuracy:
| Feature | Description | Benefit |
|---|---|---|
| Real-Time Token Streaming | Server-Sent Events (SSE) token pipeline | Sub-50ms first-word response with typewriter effect |
| SHA-256 Deduplication | Cryptographic fingerprinting on every record | Prevents duplicate records from consuming quota slots |
| Exact Citation Grounding | Every answer cites exact document names | Complete transparency and zero hallucination risk |
| In-Memory Query Cache | High-speed 60s memory caching | Returns instant 0ms responses for repeat user queries |
Free Tier Quotas & Fair Use Limits
Every developer account receives generous, permanently free quotas designed for websites, SaaS documentation, and internal support bots:
| Platform Resource | Free Tier Limit | Enforcement & Notes |
|---|---|---|
| Custom Knowledge Bases | 2 Active RAGs | Hard quota per user. Delete an old RAG to create a new one. |
| Records Capacity | 25 Records per RAG | Ideal for FAQs, policies, product catalogs, and guides. |
| Requests Rate Limit (RPM) | 10 Queries / Minute | Sliding window rate protection against spam. |
| Daily Quota (RPD) | 100 Queries / Day | Resets daily at 00:00 UTC. |
| File Upload Size | 10 MB per File | Supports JSON (.json), Plaintext (.txt), and Markdown (.md). |