Docs: Echo People | Oceanic AI

Technical Summary

Echo People is a specialized RAG (Retrieval-Augmented Generation) platform designed to capture, index, and retrieve knowledge about people, their expertise, projects, and relationships. It powers AI avatars with organizational context, enables natural language searches across professional networks, and integrates seamlessly with Slack, Teams, and communication platforms.

Core Capabilities

People Knowledge Graphs: Automatically builds relationship networks and expertise maps from organizational data
Contextual Search: Find experts by skills, projects, or relationships with semantic understanding
Avatar Training: Powers Porpoise AI Interviewers with pre-trained organizational knowledge
Chat UI SDK: Embeddable conversational interface for any application
Multi-Channel Integration: Native connections to Slack, Teams, Email for invitation workflows

Performance Metrics

Operation	Target Latency	Accuracy	Scale
Semantic Search	< 200ms	MRR ≥ 0.80	10K+ people
RAG Query	< 2s (p95)	Context recall 90%+	100K+ docs
Document Ingestion	100 docs/min	99%+ success	Multi-format

Architecture Overview

┌─────────────────────────────────────────────────────────┐
│               ECHO PEOPLE RAG PLATFORM                  │
├─────────────────────────────────────────────────────────┤
│                                                         │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐             │
│  │  People  │  │ Document │  │Knowledge │             │
│  │  Index   │  │ Ingestion│  │  Graph   │             │
│  └────┬─────┘  └────┬─────┘  └────┬─────┘             │
│       │             │             │                     │
│       └─────────────┼─────────────┘                     │
│                     │                                   │
│  ┌──────────┐  ┌────▼─────┐  ┌──────────┐             │
│  │Semantic  │  │   RAG    │  │  Chat    │             │
│  │  Search  │  │  Engine  │  │   UI     │             │
│  └──────────┘  └──────────┘  └──────────┘             │
│                                                         │
└─────────────────────────────────────────────────────────┘
                       ▲
                       │
        ┌──────────────┴──────────────┐
        │    DATA SOURCES             │
        ├─────────────────────────────┤
        │ • HR Systems (profiles)     │
        │ • Project Databases         │
        │ • Slack/Teams Activity      │
        │ • Resume/CV Documents       │
        └─────────────────────────────┘

Core Components

People Index

Structured profiles with expertise mapping

• Skills and certifications
• Project history
• Relationship networks
• Real-time updates

Document Ingestion

Multi-format document processing pipeline

• PDF, DOCX, TXT, MD
• Chunking and embedding
• Metadata extraction
• Batch processing

Knowledge Graph

Entity relationships and organizational structure

• Person-to-person links
• Project collaborations
• Skill taxonomies
• Graph traversal queries

Semantic Search

Vector-based expertise discovery

• Natural language queries
• Hybrid search (dense + sparse)
• Relevance ranking
• Sub-200ms retrieval

RAG Engine

Context-aware response generation

• Multi-document synthesis
• Source attribution
• Context relevance scoring
• LLM routing (GPT-4/Claude)

Chat UI SDK

Embeddable conversational interface

• React component library
• Customizable themes
• Streaming responses
• Multi-tenant support

Technology Stack

Backend

• FastAPI (Python)
• LangChain framework
• Celery task queue
• PostgreSQL (metadata)
• Redis (caching)

Vector Storage

• Qdrant (primary)
• pgvector (alternative)
• OpenAI embeddings
• HNSW indexing
• Cosine similarity

Integration

• Slack SDK
• Microsoft Graph API
• OAuth 2.0
• Twilio API
• Webhook handlers

Echo People RAG Platform