Core Concepts
Understanding the fundamental concepts behind the Ontbo API
API Architecture Overview
The Ontbo API is built around three core concepts that work together to provide intelligent context management for your AI applications. Understanding these concepts is essential for effectively using the API.
Overall Workflow
Understanding the complete workflow of how the Ontbo API processes and manages conversation data from ingestion to retrieval.
Learn About the Overall WorkflowProfiles and Scenes
The foundation of the Ontbo API is built on profiles and scenes, which organize and structure your conversation data for effective context management.
Profiles
Profiles represent individual users or entities in your system. Each profile contains a collection of facts and context extracted from conversations.
- • Unique identifier for each user
- • Contains extracted facts and context
- • Can be queried for relevant information
- • Automatically updated with new conversations
Scenes
Scenes represent individual conversations or interaction sessions. They contain the raw conversation data and are associated with specific profiles.
- • Individual conversation sessions
- • Contains message history
- • Associated with a specific profile
- • Can be queried for specific information
Retrieval Modes
The API offers multiple retrieval modes to query your context database, each optimized for different use cases and performance requirements.
FULL_DATA
Most Precise, SlowestPerforms an extensive search on all profile facts. Best for comprehensive analysis but may be slower for real-time applications.
MULTI_HOP
BalancedMakes multiple sub-queries to answer complex questions. Good balance between accuracy and performance for most use cases.
SINGLE_HOP
FastSimple call to the facts database. Fastest option for straightforward queries that don't require complex reasoning.
VECTOR_SEARCH
Fastest, SemanticUses embeddings to find semantically similar content. Fastest option and great for finding conceptually related information.
Profile Update Policies
Control how and when your profiles are updated with new conversation data. Understanding these policies helps you optimize performance and ensure data freshness.
Automatic Updates
Profiles can be automatically updated when new conversation data is added.
- • Set
update_now=truewhen adding text - • Profile facts are immediately processed
- • Best for real-time applications
Manual Updates
Control when profile updates occur for better performance management.
- • Add conversation data without immediate processing
- • Trigger updates manually when needed
- • Monitor update status and progress
Update Status Monitoring
Track the status and progress of profile updates in real-time.
- • Check if updates are WORKING or IDLE
- • Monitor progress percentage (0.0 to 100.0)
- • Stop updates if needed
Ready to Explore the API?
Now that you understand the core concepts, explore the interactive API documentation to see these concepts in action.