Prompt Engineering: Personalized Voice Emulation and Agentic Workflow Setup

Framework Overview

• Core Technique: Persona-based Style Mirroring (Personalization) & External Knowledge Retrieval
• Target LLMs: Optimized for ChatGPT Work; compatible with advanced reasoning models able to analyze external documentations like Gmail or Google Drive.
• Primary Use Case: Automating content creation that mimics a specific human voice while utilizing custom 'knowledge brains' or specialized toolsets.

The Prompt Template / Configuration Logic

[System Instructionset]
1. Access Source Data: {Gmail | Google_Drive | Slack | SharePoint}
2. Analyze Pattern: [Identify preferred phrases/formulations], [monitor capitalization patterns], [detect signature styles].
3. Apply Constraint: Execute all future text generation under the rule "Write in {{user_voice}}".

Execution Workflow & Rules

1. Step 1: Build an external knowledge base using methods such as Second Brain, Telegram chats, or direct file ingestion via cloud services.
2. Step 2: Configure agentic skills by providing access to personal data sources (e.g., email and documents).
3. Step 3: Enable Personalization settings within Settings -> Personalization -> Writing style(s) if available in your model version.
4. Step 4: Deploy agents for specialized tasks including web design assistance, site management, presentation building, and digest collection through automated scraping or writing tools.

Behavioral Tuning & Anti-Hallucination

• Constraint - Style Integrity: Ensure that any output generated must adhere strictly to identified characterizations like specific way of signing off messages and phraseology choice ("writing voice").
• Contextual Grounding: Use dedicated guides/bases ("Second Brain") rather than general training weight alone to ensure accuracy when performing niche professional tasks.
• Tool Constraint: When acting as a designer or website manager, the AI should follow instructions specifically tailored to those domains (Design assistant / Site maintenance logic), preventing generic responses lack appropriate technical tone.

Bottom Line: By integrating external knowledge bases with personalized writing styles via cloud integrations, you transform an LLM from a generalist into a highly customized personal agent capable of producing content indistinguishable from human input while maintaining high factual density.

! DYOR (Do Your Own Research)