DocumentInsight.ai Overview & Key Features
Your guide to know your way around the DocumentInsight.ai web application.
What is DocumentInsight.ai?
DocumentInsight.ai is a web-based application that leverages generative AI to help you search, summarize, and extract insights from your documents quickly and efficiently.
Our most prominent feature is Alerts to automatically monitor incoming emails or documents and get notified when relevant information is detected.
Key Concepts
LLM (Large Language Model)
LLM stands for Large Language Model. Examples include ChatGPT, Gemini, and Groq. These are AI models trained on massive amounts of text and visual media to learn language patterns and generate or complete text in a human-like way.
RAG (Retrieval-Augmented Generation)
RAG is a method for AI to retrieve real information and use it to provide better answers. DocumentInsight.ai is a specialized RAG application.
Analogy: If information was a field of books in the dark, plain LLMs use a single powerful light projector, common RAG uses targeted lasers on specific spots, and our approach uses many small agents with flashlights to cover the entire field.
Learn more
Key application concepts
Workspace
Your personal area where all files, folders, and snapshots are stored.
Folders
This application uses Folders & Subfolders as the basic organization block for data and queries. Files are stored in folders, queries are run on folders, and snapshots are created from folders.
Using this concept correctly, to organize files by topic in nested groups, will result in easy retrieval and targeted insights.
Original Files vs Extracted Text
Original files are securely stored and not used directly in queries. Upon upload, text is extracted or transcribed automatically. Extracted text is searchable and used in queries. You can always download the original file.
Unique Features
Smart Snapshots
Smart Snapshots let you create custom insights from documents in a folder using an AI prompt. These refined "views" enable faster, more revealing future queries.
- Appear as a dropdown in the chat area
- Selecting a snapshot uses it to find relevant documents
- "Use exclusively" mode runs queries only on the snapshot content for the fastest, most focused results
- Settings checkboxes do not affect this mode
Alerts
Alerts are an automated feature that uses AI to monitor your incoming emails or documents for specific information you define.
- Set up Alerts to watch for keywords, questions, or criteria in emails or uploaded files
- Get notified only when relevant information is detected—saving you time and reducing noise
- Supports scanning attachments (PDF, DOCX, TXT, images, audio) for your criteria
- Common use cases: compliance tracking, resume screening, monitoring newsletters, project updates, and more
- Fully private: only processes what you choose to forward/upload, and you can delete data at any time
- See FAQ: What are Alerts? for more details
GPTs Hybrid App
GPTs Hybrid App: A chat interface built on OpenAI's "custom GPTs" module, available to all subscription tiers. Securely linked to your workspace, it allows you to:
- List and find files/folders by name or content
- Get quick summaries with word count and download link
- Execute natural language queries for data and insights
- Compare files by content
Application Features & Controls
- Language Control: Choose from 13 output languages, or select "Auto" to infer from file content.
- Clarifying the Query: By default, the assistant checks if your query is clear before answering, saving credits and improving results.
- Auto-selecting Documents: The assistant picks the most relevant files for your prompt (under the selected folder), making queries faster and more efficient. This can be changed in settings.
- Adjusted Query: The assistant adapts your query to each document, ensuring better, context-aware answers. This can be toggled in settings.
-
System Prompt: A System Prompt is a top-level instruction guiding AI behavior. The default is:
"You must focus on providing evidence-based answers. For each query: 1. Extract Quotes. Include as many relevant quotes from the context. 2. Group related quotes to comprehensively support the answer. 3. Answer the query fully. 4. Embed the source links within your answer after every major group of quotes."
You may customize this prompt as needed, in settings.
- Auto Upload: When enabled in settings, files selected or dragged into the upload area will upload automatically after a 5-second countdown.
- Chat History: A running log of your chats with the assistant. Chats are automatically named when possible.
-
“Go Faster – Go Deeper” Slider:
This slider controls how deep (more thorough, slower, uses more credits) or how fast (less thorough, faster, uses fewer credits) each query runs.
- The slider position corresponds to three checkboxes in the settings panel: Clarify before answering, Skip auto-selecting documents, and Adjust query to each document.
- Checking/unchecking these boxes moves the slider and vice versa.
- Folder Content Summary: An AI-generated combined overview of everything inside a folder. It is available on the folder context menu.
Need More Help?
Try our Frequently Asked Questions page and our Instruction Videos page.
If you have further questions or need support, please contact our team through the contact page.