Hug AI
What?
Hug AI is a “human-governed AI automated management framework” for software projects, based on the “Docs-as-Code” philosophy. Its purpose is to integrate artificial intelligence agents into management and development processes while always maintaining human supervision and governance. The methodology uses exclusively text files (Markdown, YAML, JSON) organized in folders, facilitating traceability, transparency, and collaborative control.
What For?
To automate and improve software development management through AI, ensuring that critical decisions and validations always involve human intervention.
To create a transparent, audited, and versioned system of agent definitions, prompts, and AI models.
To facilitate asynchronous and documented collaboration, reducing the need for meetings and calls.
To enable the integration of automated workflows where AI operates according to clear rules and under supervision.
To foster a culture of effective governance, where AI assists but does not replace human responsibility.
To ease the onboarding of new members and the long-term maintenance of the project with living, updated documents.
Why?
The increasing incorporation of AI in development processes requires ensuring that its results are reliable, reviewable, and supervised to avoid errors or biases.
The text-only approach allows any member to access, understand, and modify configurations without complex tools.
Human governance protects quality, safety, and ethics, preventing uncontrolled automated decisions.
The methodology responds to the need for agile, flexible, and documented development, adaptable to remote and distributed teams.
Using text files and modular structures facilitates integration with other tools and automations.
It allows leveraging the advantages of AI for analysis, generation, and support, without losing the traceability and control inherent in good engineering practices.
How?
Fully textual and modular, with clear structures by areas (AI management, documentation, product, project, tools, workflows).
Each AI agent, prompt, model, and workflow is defined through readable and versionable files.
It uses YAML for structured data (configurations, metadata) and Markdown for narrative content and prompts.
It allows defining AI agent roles, their capabilities, behaviors, and explicit governance rules (for example, when to escalate to a human).
The system promotes asynchronous collaboration, avoiding unnecessary meetings and ensuring every decision and progress is documented.
It supports integration through workflows that combine AI actions and human validation.
Configurations, rules, and processes are documented and reviewed to ensure quality and compliance with internal policies.
Where?
The framework is designed to be implemented on a shared repository or workspace (e.g., Google Drive, cloud file system, GitHub).
The folder structure is housed under a root directory hugai/, with subfolders for each module (e.g., hugai/ai-management, hugai/project-management).
Its design is independent of specific platforms, aiming for compatibility with any system that handles text files and version control.
It is applicable to distributed, remote, or hybrid teams, where asynchronous communication and document-based governance are key.
When?
initial planning through delivery and maintenance.
Its use is especially valuable when:
The goal is to integrate AI to support tasks without losing human control.
Working in distributed teams with asynchronous communication.
High standards of documentation, traceability, and governance are required.
An agile system is desired without relying on constant platforms or meetings.
It can be adopted from the start of the project or implemented in existing projects to improve management.
Who?
Sebastián Larrauri: Consultant and professional with extensive experience in technology project management, digital transformation, and agile methodologies, with a focus on innovation and continuous improvement.
José Luis Reartes: Software engineer and developer with experience in open-source projects, AI integration, and process automation.
Adaptabilidad y aplicaciones más allá del desarrollo de software
Aunque HugAI está inicialmente enfocado en la gestión y automatización de proyectos de desarrollo de software, su metodología basada en documentos de texto, gobernanza humana y agentes IA es altamente flexible y adaptable a otros tipos de proyectos o áreas de gestión que involucren documentación estructurada y colaboración asíncrona.
Producción audiovisual y guionismo:
para gestionar la creación de guiones, desarrollo de personajes, planificación de escenas, revisión de diálogos y coordinación del equipo creativo, manteniendo versiones, feedback y workflows automatizados con IA (ej. un agente que analiza la coherencia del guion o sugiere mejoras narrativas).
Gestión editorial y creación de contenido
organización de publicaciones, edición de artículos, control de versiones de textos, revisión automatizada con IA y asignación de tareas entre redactores y editores.
Administración y gestión empresarial
ocumentación de procesos internos, políticas, seguimiento de tareas, generación de reportes, auditorías y workflows de aprobación, integrando IA para análisis de riesgos o detección de anomalías.
Proyectos de investigación o académicos
manejo de bibliografía, planificación de experimentos, generación y revisión de documentos científicos, seguimiento de hitos y resultados con trazabilidad completa.
Gestión de eventos o campañas
coordinación de tareas, agendas, comunicaciones y resultados, con soporte de IA para resúmenes, recordatorios automáticos y control de calidad.
Procesos Versionados
HugAI sugiere que cualquier proceso que dependa de documentos versionados, colaboración distribuida y toma de decisiones con supervisión humana puede beneficiarse de esta metodología. La modularidad y claridad de los archivos permiten adaptar agentes IA y workflows a las necesidades específicas de cada contexto, asegurando transparencia, control y eficiencia.