Chat with your documents, images, videos, and audio files • Get live news and financial data • Generate images • Remember important conversations
Pixelbot is an AI agent that can:
- 📄 Answer questions about your uploaded files (PDFs, images, videos, audio)
- 🔍 Search and retrieve information from news and financial APIs
- 🎨 Generate images from text descriptions
- 🧠 Remember important snippets in a searchable memory bank
- 🎭 Adapt its personality with custom personas
Built on Pixeltable - open-source AI data infrastructure that handles multimodal data processing, vector search, and workflow orchestration automatically.
flowchart TD
%% User Interaction
User([User]) -->|Upload Files| SourceTables["📁 File Storage<br/>Documents, Images, Videos, Audio"]
User -->|Ask Question| MainWorkflow[🤖 AI Workflow]
User -->|Generate Image| ImageGen[🎨 Image Generation]
User -->|Save Memory| MemoryBank[🧠 Memory Bank]
%% File Processing & Search
SourceTables --> AutoProcess[📊 Auto Processing<br/>Chunking, Transcription, Thumbnails]
AutoProcess --> SearchIndexes[🔍 Search Indexes<br/>Text & Image Embeddings]
%% Main AI Workflow
MainWorkflow -->|1. Choose Tools| LLM1[🧠 Claude 3.5<br/>Tool Selection]
LLM1 -->|2. Execute| Tools[🛠️ External APIs<br/>News, Finance, Search]
SearchIndexes -->|3. Find Context| ContextAssembly[📋 Context Assembly]
Tools -->|Tool Results| ContextAssembly
MemoryBank -->|Relevant Memories| ContextAssembly
ContextAssembly -->|4. Final Answer| LLM2[🧠 Claude 3.5<br/>Response Generation]
LLM2 -->|5. Response| User
%% Image Generation Flow
ImageGen -->|Prompt| DALLE[🎨 DALL-E 3]
DALLE -->|Generated Image| User
%% Memory & History
LLM2 -->|Important Info| MemoryBank
MainWorkflow -->|All Conversations| ChatHistory[💭 Chat History]
ChatHistory --> SearchIndexes
%% Styling
classDef user fill:#E3F2FD,stroke:#1976D2,stroke-width:2px
classDef storage fill:#F3E5F5,stroke:#7B1FA2,stroke-width:2px
classDef ai fill:#FFF3E0,stroke:#F57C00,stroke-width:2px
classDef process fill:#E8F5E8,stroke:#388E3C,stroke-width:2px
classDef search fill:#FFF8E1,stroke:#FBC02D,stroke-width:2px
class User user
class SourceTables,MemoryBank,ChatHistory storage
class LLM1,LLM2,DALLE ai
class AutoProcess,ContextAssembly,MainWorkflow,ImageGen process
class SearchIndexes,Tools search
Key Concept: Everything runs as declarative workflows in Pixeltable. When you upload a file or ask a question, Pixeltable automatically handles data transformations, embeddings, API calls, and LLM orchestration based on the schema defined in setup_pixeltable.py.
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install -r requirements.txtCreate a .env file:
# Required
ANTHROPIC_API_KEY=your_claude_key_here
OPENAI_API_KEY=your_openai_key_here
MISTRAL_API_KEY=your_mistral_key_here
# Optional
NEWS_API_KEY=your_news_api_key_here
# Run locally without login
AUTH_MODE=localpython setup_pixeltable.pypython endpoint.pyVisit http://localhost:5000 and start chatting!
- 🤖 Smart File Processing: Upload documents, images, videos, or audio and ask questions about them
- 🔧 External Tools: Get live news, stock prices, and web search results
- 💾 Persistent Memory: Save important information that carries across conversations
- 🎨 Image Generation: Create images with DALL-E 3 integration
- 👤 Custom Personas: Create different AI personalities for different use cases
- 📱 Clean Interface: Modern web UI with file management and conversation history
| Type | Formats | Features |
|---|---|---|
| Documents | PDF, TXT, MD, HTML | Full text search, automatic chunking |
| Images | JPG, PNG | Visual similarity search, thumbnail generation |
| Videos | MP4, MOV, AVI | Frame extraction, automatic transcription |
| Audio | MP3, WAV, M4A | Speech-to-text, searchable transcripts |
Built with ❤️ using Pixeltable - The AI Data Infrastructure