Distilly is a person-modeling framework that transforms source materials into reusable Person Profiles for AI agents and compatible bots.
The tool addresses the problem of preserving and reusing someone's experience, judgment, communication style, and decision-making patterns when they leave a role, relationship, or are no longer accessible. It works by ingesting messages, documents, interviews, and public sources, then distilling these materials into a source-grounded Person Profile that captures observable patterns without claiming to clone the person. The resulting profile can be packaged as an Agent Skill and installed in supported hosts, allowing agents to invoke the distilled knowledge and perspective.
Distilly suits scenarios where you want to retain institutional knowledge from departing colleagues, maintain connection with distant relationships, or model the thinking of public figures and authors. The tool explicitly does not claim to replicate a person's identity, only to ground profiles in observable experience and expression. It works with diverse source material types and supports installation across multiple agent platforms including Claude Code, Hermes, OpenClaw, Codex, DeepSeek Harness, and others. The project maintains a community gallery where users contribute and share Person Profile skills.
Development shows sustained momentum with a published technical report documenting the underlying methodology. The project has expanded its scope beyond the original colleague-focused use case to encompass family, friends, partners, public figures, and fictional characters. Community contribution is active, with a growing gallery of shared skills and documented support across multiple agent platforms and frameworks. The project maintains multilingual documentation and an active Discord community for discussion and collaboration.
