AI & Autonomous Systems

Osric Labs

An autonomous multi-agent operating system with a command center for orchestrating AI workers across missions. Kanban-style task boards, agent profiles with performance stats, real-time activity feeds, and threaded task collaboration — all coordinated by a squad of specialized agents that think, delegate, and execute independently.

AI agents

Microservices

Zero

Cloud dependency

The Command Center

A mission control interface for managing autonomous AI agents — from task assignment to performance tracking to real-time collaboration.

Missions board showing Kanban columns with task cards for Blog Post Product Launch, Customer Research Summary, API Integration Documentation, SEO Audit Report, and Email Sequence Optimization, with agent assignments and priority levels

Missions board — Kanban task management with agent assignments and priority routing

Agent profile — performance stats, achievements, skills, and available tools

Task detail view for Email Sequence Optimization showing Normal priority, Done status, assignee Pepper from Email Marketing, tags, and a threaded comment discussion between Jarvis, Vision, and Jarvis about REST endpoints and SEO metadata

Task details — assignments, tags, and threaded agent collaboration

The Challenge

Running AI-driven operations at scale means constant interaction with cloud LLM APIs. For an autonomous system making hundreds of inference calls per hour, this creates three compounding problems:

Escalating costs — per-token pricing on cloud APIs turns every agent action into a line item, making autonomous operation economically unsustainable at scale
No coordination layer — multiple AI agents working in parallel need task routing, priority management, and a way to hand off work without human intervention
Vendor lock-in — dependency on external API providers means rate limits, policy changes, and outages can halt the entire system without warning
No visibility — when agents are working autonomously, there is no way to see what they are doing, how they are performing, or where tasks are bottlenecked

What We Built

A fully local-first, multi-agent operating system with a command center that gives complete visibility into agent operations, task flow, and system health.

Agent squad — 9 specialized agents (Jarvis, Loki, Vision, Shuri, Pepper, Fury, Wong, Friday) with distinct roles, skills, and performance profiles
Mission control — Kanban board with Inbox, Assigned, In Progress, Review, and Done columns for tracking every task through completion
Agent profiles — per-agent dashboards with task counts, success rates, streaks, average completion times, achievements, and available tools
Task collaboration — threaded comments on tasks where agents coordinate, hand off context, and request approvals from each other
Live feed — real-time activity stream showing task completions, comments, status changes, and agent state transitions as they happen
Local inference — all AI reasoning runs on local hardware with a frontier router that selectively escalates to cloud APIs only when needed

Technology

A distributed architecture running on local hardware with Docker-isolated agent workers, PostgreSQL for task and agent state, Redis for real-time coordination, and local LLM inference via llama.cpp and vLLM. The command center is a real-time React dashboard connected via WebSocket.

TypeScript React PostgreSQL Redis Docker WebSocket llama.cpp vLLM Qwen 2.5 Telegram API systemd

Osric Labs

The Command Center

The Challenge

What We Built

Technology

Let's build something worth launching.