Generic AI Outreach Is a Context Problem

I have analyzed the two examples of AI-generated outreach messages provided in the source material. The first message, Output A, was generated using a prompt that simply asked the model to create a personalized opening line for a company hiring salespeople. The second message, Output B, was created using a more detailed prompt that included context about the company posting three SDR job listings, a specific Ideal Customer Profile (ICP) rule, a persona of a VP of Sales focused on ramp time and rep replication, a signal hypothesis, and a proposed proof point.

The key difference between the two outputs is the level of context and specificity provided in the prompt. Output A, while still personalized to some extent by referencing the hiring, lacks a clear connection to the recipient's current concerns and needs. In contrast, Output B is tailored to the VP of Sales persona and explicitly addresses the issue of ramp time and rep replication by suggesting a solution based on the company's recent SDR hiring.

This analysis demonstrates that the quality of AI-generated outreach is heavily dependent on the context and specificity provided in the prompt. Without a well-defined context assembly step that includes the ICP rule, persona priorities, signal hypothesis, and proof mapping, the AI model is likely to generate generic and less effective outreach messages.

The source material emphasizes that the model itself does not inherently produce better or worse results; rather, the issue lies in the quality of the context provided to the model.