Employee Clone

Problem

Personal know-how often sits scattered across chats, docs, emails, and past decisions. General-purpose AI can answer, but it usually cannot reproduce a specific engineer's, salesperson's, or support agent's judgment style, communication pattern, and operational boundaries. This skill targets scenarios where an individual's working method needs to be preserved: it turns authorized materials into a reusable digital clone, then uses that clone to provide suggestions in customer support, technical troubleshooting, sales scripts, and product analysis.

How It Works

Instead of relying on an external knowledge base, it builds a local profile from configuration files. Creation can be guided through quick, standard, or deep Q&A flows. It can import documents, chat records, emails, and decision cases, then model four dimensions: domain knowledge, communication style, decision preferences, and common phrasing. After training, you can use dialogue to simulate how the person would respond, scripts to generate reply suggestions, and decision to reason through “what would they choose?” It also includes consistency validation by comparing real and cloned outputs with a fidelity score, supports ongoing learning from new cases, and allows version rollback. A trained clone can be packaged as a standard SKILL and installed into another agent.

Boundaries

The skill is advisory only: it does not perform external actions and does not make final decisions for the user. Materials must be authorized by the person being cloned, and cloning another person is not allowed. The described scripts only process local JSON configuration, make no network requests, execute no dynamic code, and invoke no external commands. It is suitable for style-assisted drafting and personal workflow support, not for approvals, transactions, releases, or operations where a responsible human must act.