Show HN: Built software to orchestrate AI and human loops (github.com)

A new software called cognitive-firm has been introduced, designed to streamline the integration of AI agents and human roles within organizations. This "governance kernel" enables principals to manage persistent AI agents with clear mandates, budget limitations, and approval processes, distinct from conventional agent frameworks or SDKs. It aims to facilitate structured human-agent collaboration by establishing roles, maintaining audit trails, and documenting interactions—all while allowing agents to operate autonomously within predefined boundaries. This unique approach prioritizes accountability and role clarity, ensuring every task has an identifiable human or agent responsible for results and records. The significance of cognitive-firm for the AI/ML community lies in its innovative method for governing AI involvement in organizational tasks. Unlike existing frameworks that emphasize either agent autonomy or complete human oversight, cognitive-firm balances the two, promoting efficiency and accountability in workflows. With built-in protocols for human-agent and agent-agent interactions, it provides a robust foundation for persistent operational roles. Furthermore, the software is crafted to be reusable and easily inspectable, which benefits researchers and small teams looking to develop more accountable AI systems without extensive overhead. This opens new avenues for organizational learning and improved collaboration between human and machine actors.