Features Overview

SeanceAI is packed with features designed to make conversations with historical figures feel authentic, engaging, and educational.

Core Features

Seance Mode

One-on-one conversations with any of 60+ historical figures. Ask questions, discuss ideas, and explore their unique perspectives in intimate dialogue.

Dinner Party Mode

Multi-figure discussions with 2-5 historical guests. Watch as they debate, agree, disagree, and respond to each other authentically.

60+ Historical Figures

Converse with legends from Ancient World, Renaissance, 19th Century, and Modern Era. Each figure meticulously crafted with unique personality and knowledge.

Streaming Responses

Real-time streaming using Server-Sent Events (SSE) for natural, flowing conversations. See responses appear word-by-word as the AI generates them.

Authentic Historical Personas

Era-Appropriate Knowledge

Each historical figure knows only what they would have known during their lifetime:

# From app.py - System prompt ensures historical accuracy
system_prompt = get_system_prompt(figure_id)
messages = [{"role": "system", "content": system_prompt}]

Figures don’t have knowledge of events after their death, creating authentic historical perspectives.

Unique Personalities

Speaking styles match each figure’s documented communication patterns
Vocabulary appropriate to their time period and profession
Personality traits based on historical research
Genuine reactions to modern concepts they wouldn’t understand

Historical Figures Available

Ancient World

Cleopatra, Julius Caesar, Socrates, Plato, Aristotle, Archimedes, Sun Tzu, Confucius, Alexander the Great, and more.

Renaissance Era

Leonardo da Vinci, Michelangelo, Galileo Galilei, Nicolaus Copernicus, Johannes Kepler, William Shakespeare, and more.

19th Century

Abraham Lincoln, Charles Darwin, Nikola Tesla, Marie Curie, Napoleon Bonaparte, Ada Lovelace, Florence Nightingale, and more.

Modern Era (20th Century)

Albert Einstein, Alan Turing, Stephen Hawking, Richard Feynman, Carl Sagan, Rosalind Franklin, and more.

Intelligent Conversation Features

Contextual Suggestions

AI-generated follow-up questions based on the current conversation:

# From app.py - Generates contextual suggestions
@app.route('/api/suggestions', methods=['POST'])
def api_suggestions():
    # Analyzes conversation history and last response
    # Returns 2-3 relevant follow-up questions
    return jsonify({"suggestions": suggestions})

Suggestions are tailored to each conversation, helping you explore deeper topics or change direction naturally.

Conversation History

SeanceAI maintains context throughout your conversation:

# From app.py - History management
MAX_HISTORY = 20  # Keep last 20 messages for context

# Build messages with history
for msg in history[-MAX_HISTORY:]:
    messages.append({
        "role": msg.get("role", "user"),
        "content": msg.get("content", "")
    })

Maintains context for natural, flowing dialogue
References previous exchanges in responses
Remembers topics discussed earlier in the conversation

Save & Export

LocalStorage persistence - conversations automatically saved in browser
Export to text - download conversations for later reference
Resume anytime - return to saved conversations

Advanced AI Capabilities

Multi-Model Support

SeanceAI integrates with OpenRouter to provide access to multiple AI models:

# From app.py - Available models organized by tier
AVAILABLE_MODELS = [
    # Swift tier - fast and responsive (free models)
    {"id": "google/gemma-3-12b-it:free", "name": "Gemma 3 12B", "tier": "swift"},
    {"id": "meta-llama/llama-3.3-70b-instruct:free", "name": "Llama 3.3 70B", "tier": "swift"},
    
    # Balanced tier - good mix of speed and capability
    {"id": "openai/gpt-4o-mini", "name": "GPT-4o Mini", "tier": "balanced"},
    {"id": "anthropic/claude-3.5-haiku", "name": "Claude 3.5 Haiku", "tier": "balanced"},
    
    # Advanced tier - most capable models
    {"id": "anthropic/claude-sonnet-4", "name": "Claude Sonnet 4", "tier": "advanced"},
    {"id": "openai/gpt-4o", "name": "GPT-4o", "tier": "advanced"},
]

Three Capability Tiers:

Swift

Fast, responsive free models. Great for quick conversations. Includes Gemma 3 and Llama 3.3.

Balanced

Mix of speed and capability. Includes GPT-4o Mini and Claude Haiku.

Advanced

Most capable models for complex discussions. Includes GPT-4o and Claude Sonnet 4.

Automatic Fallback System

Intelligent retry logic handles rate limits gracefully:

# From app.py - Fallback model handling
FALLBACK_MODELS = [
    "google/gemma-3-27b-it:free",
    "google/gemma-3-4b-it:free",
    "meta-llama/llama-3.3-70b-instruct:free",
    "meta-llama/llama-3.1-405b-instruct:free",
]

MAX_RETRIES = 3
RETRY_DELAYS = [2, 5, 10]  # Exponential backoff

How it works:

Tries your selected model
If rate-limited, waits and retries with exponential backoff
After max retries, automatically switches to fallback models
Ensures you always get a response

If all models are rate-limited, you’ll see a helpful message to wait or select a different model.

Streaming Responses

Real-time response generation using Server-Sent Events:

# From app.py - Streaming endpoint
@app.route('/api/chat/stream', methods=['POST'])
def api_chat_stream():
    return Response(
        stream_llm(messages, model),
        mimetype='text/event-stream',
        headers={
            'Cache-Control': 'no-cache',
            'Connection': 'keep-alive',
            'X-Accel-Buffering': 'no'
        }
    )

Word-by-word generation for natural conversation feel
Lower latency - see responses immediately
Better user experience - no waiting for complete response

Dinner Party Features

Multi-Figure Conversations

# From app.py - Dinner party endpoint
@app.route('/api/dinner-party/chat', methods=['POST'])
def api_dinner_party_chat():
    guest_ids = data.get('guests', [])
    
    # Validate 2-5 guests
    if len(guest_ids) < 2 or len(guest_ids) > 5:
        return jsonify({"error": "2-5 guests required"})
    
    # Build dinner party prompt with all guest personalities
    system_prompt = get_dinner_party_prompt(guest_ids)

Features:

2-5 historical figures in one conversation
Each responds authentically to your questions
Figures interact with each other - debates, agreements, discussions
Dynamic dialogue that feels like a real dinner party

Curated Combinations

Pre-selected guest combinations for great discussions:

# From figures.py - Curated dinner party combinations
CURATED_COMBOS = [
    # Scientists & Mathematicians
    ["einstein", "newton", "tesla"],
    
    # Philosophers
    ["socrates", "plato", "aristotle"],
    
    # Renaissance Minds
    ["davinci", "galileo", "michelangelo"],
    
    # And more...
]

User Interface

Museum-Themed Design

Elegant dark UI with gold accents
Beautiful SVG portraits for every figure
Smooth animations and transitions
Responsive design - works on all devices

Search & Filter

Search by name - quickly find specific figures
Filter by era - browse by time period
Biographical information for each figure
Starter questions to help begin conversations

Mobile Optimized

Fully responsive layout
Touch-friendly interface
Optimized for tablets and phones
Same great experience across devices

Health Monitoring

# From app.py - Health check endpoint
@app.route('/api/health')
def api_health():
    return jsonify({
        "status": "healthy",
        "api_key_configured": bool(OPENROUTER_API_KEY)
    })

Use /api/health to:

Monitor service status for deployments
Check API configuration before use
Set up uptime monitoring with services like UptimeRobot

Educational Use Cases

History Education

Students can interact with figures they’re studying, asking questions and exploring historical perspectives firsthand.

Critical Thinking

Understanding different perspectives and historical context through authentic dialogue.

Language Learning

Practice conversations with figures from different eras, experiencing various speaking styles.

Creative Writing

Get inspiration from historical voices and perspectives for stories and projects.

API Integration

SeanceAI provides a REST API for programmatic access:

List All Figures

curl http://localhost:5000/api/figures

Get Single Figure

curl http://localhost:5000/api/figures/einstein

Send Chat Message

curl -X POST http://localhost:5000/api/chat \
  -H "Content-Type: application/json" \
  -d '{
    "figure_id": "einstein",
    "message": "What inspired your theory of relativity?",
    "history": [],
    "model": "google/gemma-3-12b-it:free"
  }'

List Available Models

curl http://localhost:5000/api/models

All endpoints support optional model selection via the model parameter.

Get Started

Core Features

​Features Overview

​Core Features

Seance Mode

Dinner Party Mode

60+ Historical Figures

Streaming Responses

​Authentic Historical Personas

​Era-Appropriate Knowledge

​Unique Personalities

​Historical Figures Available

​Intelligent Conversation Features

​Contextual Suggestions

​Conversation History

​Save & Export

​Advanced AI Capabilities

​Multi-Model Support

Swift

Balanced

Advanced

​Automatic Fallback System

​Streaming Responses

​Dinner Party Features

​Multi-Figure Conversations

​Curated Combinations

​User Interface

​Museum-Themed Design

​Search & Filter

​Mobile Optimized

​Health Monitoring

​Educational Use Cases

History Education

Critical Thinking

Language Learning

Creative Writing

​API Integration

​List All Figures

​Get Single Figure

​Send Chat Message

​List Available Models

Build docs developers (and LLMs) love

Features Overview

Core Features

Authentic Historical Personas

Era-Appropriate Knowledge

Unique Personalities

Historical Figures Available

Intelligent Conversation Features

Contextual Suggestions

Conversation History

Save & Export

Advanced AI Capabilities

Multi-Model Support

Automatic Fallback System

Streaming Responses

Dinner Party Features

Multi-Figure Conversations

Curated Combinations

User Interface

Museum-Themed Design

Search & Filter

Mobile Optimized

Health Monitoring

Educational Use Cases

API Integration

List All Figures

Get Single Figure

Send Chat Message

List Available Models