Structured Operational
Intelligence

Central multi-agent orchestration console. SSE streaming chat, MCP tool bridge, case context injection, and planning continuity across sessions. Auth-gated with user management.

• Multi-agent orchestration with SSE streaming
• MCP bridge exposing VPS API endpoints as tools
• Case context injection into every agent run
• User auth, admin panel, activity tracking
• Desktop + mobile + resident interfaces

Multi-jurisdiction litigation intelligence UI. Cross-border legal analysis with jurisdiction-aware framework mapping and structured document processing. Embedded within Olivia workspace.

• Multi-jurisdiction (BR, CL, INT) support
• Framework-aware legal document processing
• Integrated with Olivia auth and agent runtime

Human-guided autonomous agent platform with MCP tool orchestration. Chat-and-cards UI with SSE streaming, step-level observability, and trajectory logging for progressive workflow learning.

• Service gallery of MCP-exposed capabilities
• 11 structured step types (Say, Ask, ToolCall, Done...)
• SSE streaming with cost tracking per session
• JSONL trajectory logging for workflow learning
• React SPA + FastAPI backend

Platform nerve centre. Monitor ~30 services, control Docker containers and host processes, tail logs, run API test suites, deploy pages, and collect network telemetry — all from one panel.

• Service registry & start/stop/restart controls
• Docker + filesystem + journalctl log viewer
• API test suite (~75 endpoint definitions)
• Network telemetry capture & analysis
• VPS page deployer + endpoint mapper

Browser-based audio processing suite wrapping PyTorch torchaudio. Filter, enhance, analyze, and separate audio with 24 REST endpoints — no desktop software needed.

• Biquad filters (8 types), audio effects (8 types)
• Voice enhancement (7 types), spectral analysis (6 types)
• HDemucs source separation
• Voice activity detection + resampling
• wav2vec2 ASR via MCP server (8 tools)

Multi-tribunal jurisprudência search assistant powered by DeepSeek AI. Chat interface with intelligent field extraction, automated scraping of TJRS, TJSP, and STF, and structured result download.

• AI-assisted search form — upload a document, get fields
• Automated scraping: TJRS, TJSP, STF (Selenium)
• Chat interface with DeepSeek (v4-pro / v4-flash)
• Progress polling & structured result download
• Inteiro teor PDF download for selected results

AI-powered legal framework analyzer generator. Transforms PDFs and legal text into interactive JavaScript analyzers that detect violations in transcripts and conversations.

• Upload legal frameworks → generate JS analyzers
• Law corpus registry (BR, CL, INT jurisdictions)
• v2.4 ontology analyzer pack builder
• Multi-language UI (5 languages)
• MCP servers: law lookup + builder tools

Point it at any folder — get a fully categorized, searchable, API-ready file organization dashboard with a layered intelligence pipeline from ingestion through legal verification.

• L0–L8 pipeline: Ingest → Classify → Analyze → Verify
• SHA-256 deduplication + Jaccard near-dup detection
• LLM entity extraction & chronological narrative synthesis
• Neo4j graph + Qdrant vector knowledge base
• MCP server with 68 tools + Electron desktop app

AI Tools & Services Hub. Wraps Ollama local LLMs, Qdrant vector database, and ingestion pipelines into a unified OpenAI-compatible REST API with assistant management and prompt engineering.

• Ollama LLM wrapping + Qdrant vector operations
• OpenAI-compatible chat completions API
• Legal & transcript ingestion pipelines
• Prompt engineering lab + assistant CRUD
• 5 MCP servers (core, files, ingestion, prompt, qdrant)

Rendered shared transcript timelines. Browse and access processed audio transcript outputs with speaker diarization, timestamps, and structured analysis — served as static HTML with directory listing.

• Auto-indexed directory of rendered transcripts
• Speaker-separated with timestamps
• Structured legal analysis output from Pinocchio
• Served directly by nginx from shared storage

Chilean Spanish audio transcription and speaker diarization service. Combines OpenAI Whisper ASR with Pyannote diarization, optimized for legal and evidentiary audio analysis with Qdrant semantic search.

• Whisper ASR + Pyannote speaker diarization
• Chilean Spanish colloquial normalization
• AI transcript analysis (Claude/DeepSeek)
• Reference-guided transcription with LLM reconciliation
• Qdrant vector search + project management