Context drifts. Preferences fade. Conversations reset.
And just like that - your users churn.
Retention is won through relevance.
Shiki makes AI hyper-personalised to your customers
Transform your AI from forgetful to unforgettable with enterprise-grade memory capabilities
Capture and retain user facts, preferences, conversation tone, and context across all sessions. Never lose important details again.
Integrate in just 2 API calls. Get up and running in minutes, not hours. Optimized for speed and reliability.
Model-agnostic architecture. Local-first deployment options. Enterprise-ready security and compliance.
Handle thousands of users with isolated memory contexts
End-to-end encryption with configurable data retention
Automatically learns and adapts to user preferences
Simple integration, powerful results. Choose your preferred language and start building.
import requests
# Shiki API Configuration
API_URL = "http://127.0.0.1:8000"
# Extract and update user memories
extract_payload = {
"latest_exchange": "I just moved to Tokyo and I love sushi.",
"recent_messages": [
"Where do you live?",
"I used to live in Paris.",
"What food do you like?",
"I like Italian food."
],
"user_id": "test_user",
"top_k": 10,
"teach_preferences": True
}
# Retrieve relevant memories
get_params = {
"query": "Where do I live?",
"user_id": "test_user",
"top_k": 4
}
# Store memories
response = requests.post(f"{API_URL}/memories", json=extract_payload)
# Retrieve memories
response = requests.get(f"{API_URL}/memories", params=get_params)