Back to directory

Sico

An open-source platform for building, managing and evolving Digital Workers: structured AI labor units that combine reasoning, tool-based action and continuous learning, aimed at real production work such as business process outsourcing.

Sico

Sico defines four roles around a Digital Worker: an Employer who sets business objectives and outcome standards, an Operator who trains, monitors and improves the worker, a Developer who builds its capability structure, workflows, tools and execution environments, and the Digital Worker itself, which executes tasks through structured capabilities and continuous learning. A Digital Worker's anatomy includes a Cortex for reasoning and planning using LLMs and agent frameworks, and an Action layer for execution through domain skills, workflows and sandboxed environments.

The project grew out of large-scale operational challenges Microsoft observed internally in BPO-style workflows such as black-box testing, and reports achieving closed-loop validation for Digital Workers under continuous execution, evaluation and human supervision on real production workloads. Recent releases have added multi-tenant organization management, prepared task environments, Linux workstation sandboxes, role-based access control, scheduled tasks and multi-session chat with project context.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.