These Prompts Will 10x Your AI Output | Raj Shamani Clips Raj Shamani Clips ·
Watch on YouTube ·
Generated with SnapSummary
· 2026-05-02
Speaker introduces topic: wants to learn how to write the best prompts and create segments.
Mentions a "magic prompt formula" and desire to use AI to get better outputs.
Explains that being specific changes the kind of output you get; gives example of rewriting a simple leave-request email.
Introduces the magic prompt formula components: Role, Objective, Context, Data.
Describes "Role": pick the best person to write the email (e.g., experienced email copywriter from Ogilvy).
Defines the Objective: write an email to convince the boss to grant two days off for personal work despite having no leave balance.
Provides Context: boss is 35–45, prefers in-office presence, values consistency; tune tone to avoid upsetting him.
Notes Data: not needed here unless including specific data points or dates.
Shows the old/basic leave email example and critiques it as unconvincing.
Demonstrates emotionally sincere and then warns against overly emotional extremes as a joke.
Presents an improved emotionally sincere version of the leave email.
Suggests using multiple LLMs ("Multi") to compare outputs and shows running several models (GPT-4-style reasoning models, GBD5, Cloud4, 2.5 Pro, DeepSea/Carbon).
Observes some models are slower (reasoning models) and one model timed out/errored.
Compares subject lines and solutions across models; notes differences in responses.
Emphasizes delicate balance of respect, responsibility, and strategic communication for such emails.
Proposes creating a tool to auto-generate prompts so the speaker doesn't rewrite prompts repeatedly.
Describes building a "magic prompt generator" (a meta-prompt) using the Role/Objective/Context/Data structure.
Demonstrates using ChatGPT Projects (or similar) to store instructions so repeated prompts need not be re-entered.
Shows creating a developer/system prompt to convert user input into optimized "magic prompts."
Tests the prompt generator by asking it to create a sales email prompt to land high-paying clients (AI consulting for SMEs).
Receives generated magic-prompt-based outputs and copies/pastes them to produce final emails.
Concludes: learned the magic prompt formula, converted it into a meta-prompt via an optimizer tool, and used it to generate better emails; invites viewers to subscribe.
Summarize any YouTube video instantly
Get AI-powered summaries, timestamps, and Q&A for free.
