OTTO Documentation
Superior Intelligence Interface v3.3 — On-Premises AI
One interface to access all 454 AI models across 17 platforms. Just ask in natural language—OTTO handles the rest.
NLP Stack — all-MiniLM-L6-v2 + en_core_web_trf
Dense 384-dim sentence embeddings for domain classification. Transformer NER for financial entity extraction. Difflib typo correction. 100% accuracy on 39-query benchmark (+20.5pp vs classical TF-IDF). All inference runs on-premises — zero data sent to external APIs.
OTTO - Superior Intelligence Add-On
SBERT dense-embedding NLP (100% benchmark accuracy), en_core_web_trf transformer NER, SSE streaming, Action Cockpit, Fast Learning, and voice summarisation. One on-premises interface to all 454 AI models. Zero-knowledge architecture.
Getting Started
Quick start guide to using OTTO — on-premises setup, first query, and SDK integration
NLP Intelligence Engine
SBERT all-MiniLM-L6-v2 dense embeddings + spaCy en_core_web_trf NER. 100% accuracy on canonical benchmark. All inference on customer premises.
Streaming Intelligence (SSE)
Word-by-word answer streaming via Server-Sent Events. First token < 50ms. Full intelligence card delivered in final event.
Fast Learning
Ingest your CRM, ERP, or custom KPI data at runtime. SQLite-persisted context enriches every subsequent query with your numbers.
Voice Commands
Web Speech API multilingual voice summarisation — reads intelligence card findings aloud in 11 languages.
Action Cockpit
Execute AI-recommended actions. Real-time QUEUE → EXECUTING → DONE lifecycle with before/after outcome charts.
Zero-Knowledge Architecture
NLP engine, BIE, Monte Carlo simulation, and learning store run entirely on your premises. BrainPredict never receives your business data.
Partner SDK (v3.3)
JavaScript/TypeScript SDK. OTTOClient with streamAsk(), nlpCapabilities(), learn(), executeAction(). FastLearningClient for data-only integrations.