Skip to main content

​
All Features at a Glance

DecipherIt combines multiple AI-powered features to transform how you conduct research. Each feature is powered by specialized CrewAI agents and cutting-edge AI services.

​
Core Research Features

Deep Research

Topic-based research powered by Bright DataEnter any topic and let AI agents strategically plan data collection, discover diverse sources through global search, and synthesize comprehensive insights.
  • Web Scraping Planner creates research strategy
  • Link Collector uses search_engine to find sources
  • Web Scraper extracts content with scrape_as_markdown
  • Research Analyst synthesizes findings
  • Content Writer creates structured summaries

Multi-Source Research

Combine documents, URLs, text, and topicsSeamlessly integrate multiple input types into unified research spaces with AI-powered processing.
  • Documents: PDF, DOCX, PPTX, XLSX via MarkItDown
  • URLs: Web scraping via Bright Data MCP
  • Manual Text: Direct text input processing
  • Topics: Automated source discovery
  • All processed and integrated together

AI-Powered Summaries

Comprehensive research analysisCrewAI agents work together to create well-structured, engaging summaries from all your sources.
  • Research Analyst synthesizes information
  • Content Writer crafts engaging narratives
  • Highlights key insights and trends
  • Connects information across sources
  • Powered by Google Gemini via OpenRouter

Interactive Q&A

Chat with your research using vector searchNatural language question answering powered by semantic search through Qdrant vector database.
  • OpenAI embeddings for semantic understanding
  • Qdrant vector database for fast retrieval
  • Context-aware responses from Decipher agent
  • Chat history for follow-up questions
  • Source citations in every answer

​
Content Generation Features

Smart FAQ Generation

Automatic Q&A creation from researchAI agents analyze your research content to automatically generate relevant, insightful questions with comprehensive answers.
  • Analyzes content patterns automatically
  • Generates relevant questions
  • Creates comprehensive answers
  • Perfect for documentation
  • Ideal for study guides

Audio Overviews

Podcast-style audio summariesTransform research into engaging 4-5 minute podcast conversations with multiple AI voices.
  • Podcast Script Generator creates dialogue
  • Conversation between Michael (host) & Sarah (expert)
  • LemonFox TTS for high-quality voices
  • 800-1000 word natural conversations
  • Stored in Cloudflare R2

Visual Mindmaps

Interactive hierarchical visualizationsAI-generated mindmaps with up to 5 levels of depth to visualize research structure and relationships.
  • Content Analyzer identifies themes
  • Mindmap Creator builds hierarchy
  • Adaptive depth (2-5 levels)
  • Built with react-mindmap-visualiser
  • Interactive exploration

​
Advanced Capabilities

​
Global Web Access

Powered by Bright Data’s MCP Server - the official Model Context Protocol server for unrestricted web access:

Search Engine

Discover relevant sources globally without geo-restrictions using Bright Data’s search capabilities

Web Scraping

Extract clean, structured content from any website with anti-bot detection and residential proxies

Geo-Unrestricted

Access content from anywhere in the world regardless of location constraints

Bot Detection Bypass

Advanced techniques to avoid getting blocked during data collection

​
Vector-Powered Intelligence

Semantic search and intelligent retrieval powered by modern vector database technology:
High-performance vector database storing research embeddings:
  • Fast similarity search
  • Semantic understanding
  • Contextual retrieval
  • Scalable storage

​
Multi-Agent Architecture

DecipherIt employs specialized CrewAI crews for different tasks:
Agent: Web Scraping PlannerCreates strategic research plans and optimal data collection approaches for topics.Output: Search queries and collection strategy
Agent: Link CollectorUses Bright Data’s search_engine tool to find relevant sources globally based on research topics and search queries.Output: List of relevant URLs with titles
Agent: Web ScraperUses Bright Data’s scrape_as_markdown tool to extract and convert web content to clean, structured markdown format.Output: Clean markdown content from URLs
Agent: Research AnalystSynthesizes information from multiple sources, identifies patterns, and extracts key insights across all research material.Output: Structured research analysis
Agent: Content WriterCreates engaging, well-structured research summaries that highlight key insights, trends, and connections.Output: Comprehensive blog-style summaries
Agent: FAQ GeneratorAnalyzes content patterns to automatically generate relevant questions with comprehensive answers.Output: List of Q&A pairs
Agents: Content Analyzer + Mindmap CreatorAnalyzes research structure to identify themes and creates hierarchical mindmap visualizations with adaptive depth.Output: Interactive mindmap structure
Agents: Research Analyst + Conversation Planner + Script WriterCreates podcast-style audio overviews through multi-stage pipeline: analysis, outline, script generation, and TTS conversion.Output: Audio file with conversational summary
Agent: DecipherAnswers user questions using vector search retrieval and chat history context, providing accurate responses with source citations.Output: Contextual answers with sources

​
Processing Pipeline

​
Immediate Processing

These features run automatically when you create a notebook:
1

Source Collection

All sources (documents, URLs, text, topics) are collected and prepared for processing
2

Content Extraction

  • Documents converted to markdown via MarkItDown
  • URLs scraped via Bright Data MCP Server
  • Topics researched via Planning + Discovery + Scraping crews
3

Vector Embeddings

Content chunked and converted to embeddings using OpenAI, stored in Qdrant for semantic search
4

Research Analysis

Research Analyst synthesizes information from all sources
5

Summary Generation

Content Writer creates comprehensive, engaging summary
6

FAQ Creation

FAQ Generator analyzes content and creates Q&A pairs

​
On-Demand Generation

These features generate when you request them:

Audio

Click audio button to trigger Script Writer crew and LemonFox TTS conversion

Mindmap

Request mindmap to trigger Content Analyzer and Mindmap Creator crews

Q&A

Ask questions anytime - Decipher agent retrieves context via Qdrant search

​
Technology Stack

​
Frontend

Next.js 15

React framework with App Router for modern web applications

React 19

Latest React with concurrent features and improved performance

TypeScript 5

Type-safe development for reliable applications

Shadcn/ui

Beautiful, accessible component library with Radix UI primitives

Better Auth

Modern authentication with email/password and GitHub OAuth

Prisma

Type-safe database ORM for PostgreSQL

​
Backend

Python 3.12

Latest Python with performance improvements

FastAPI

High-performance async API framework

CrewAI

Multi-agent AI framework for complex tasks

SQLAlchemy

Python SQL toolkit and ORM

​
AI & ML Services

Bright Data MCP

Official Model Context Protocol server for web access

Google Gemini

LLM via OpenRouter for content generation

OpenAI Embeddings

Text embeddings for semantic search

LemonFox TTS

High-quality text-to-speech synthesis

Qdrant

Vector database for semantic search

MarkItDown

Document conversion to markdown

​
Infrastructure

PostgreSQL

Robust relational database for application data

Cloudflare R2

Object storage for files and audio

Docker

Containerization for deployment

​
Learn More

Deep Research

Learn how topic-based research works with Bright Data and CrewAI

Interactive Q&A

Understand vector search and conversational AI

Audio Overviews

Explore podcast generation pipeline

Mindmaps

Discover hierarchical visualization

Architecture

Deep dive into technical architecture

Self-Hosting

Deploy your own instance
All features are powered by open-source technologies and can be self-hosted. Check out the GitHub repository to explore the code!

Build docs developers (and LLMs) love

Get started for freeTalk to us