How I AIMastering ChatGPT: Advanced techniques for workplace communication and productivity | Hiten Shah
CHAPTERS
Why ChatGPT became Hiten’s daily driver (and what changed vs. Claude)
Hiten explains how his usage evolved from early OpenAI API experiments to being more Claude-heavy, then shifting back to ChatGPT. The turning point was ChatGPT’s Memory and the workflow advantages that came with it, especially for repeated, context-rich work.
Memory hygiene: temporary chats, archiving, and pruning to control what it learns
The conversation gets tactical about managing ChatGPT’s memory so it stays useful rather than noisy. Hiten shares his personal habits for limiting unwanted retention and keeping long-term personalization high-signal.
Context is everything: front-load inputs, ask for needed context, and show “what great looks like”
Hiten lays out his core philosophy: great outputs come from great inputs, and that means context, examples, and reusable structure. He emphasizes teaching the model by demonstration—using high-quality exemplars to shape consistent future results.
Demo setup: building a “Project for Projects” mindset (systems that create systems)
Before the first major demo, Hiten describes how he creates meta-projects that help him create other projects and prompts. This is positioned as a leverage tactic: invest once in a reusable setup and benefit repeatedly.
Demo: “What Would Morgan Do?” — replicating your boss for better communication
Hiten demonstrates creating a ChatGPT Project that simulates his boss Morgan’s feedback style. By uploading Morgan’s operating manual and a favored article, then generating project instructions, he uses the project to draft an effective pitch tailored to Morgan.
Iterate and codify: turning a good output into durable project instructions
Hiten highlights an important pattern: when the model produces something close-but-not-right, he pushes it to generate paste-ready instructions. When it produces something great, he captures it and turns it into a repeatable template.
Personality frameworks as “relationship context” (Enneagram, DISC, Myers-Briggs, more)
They explore enhancing AI coaching by adding personality frameworks to project context. Hiten argues these frameworks become especially powerful when combined with real artifacts and used to bridge communication differences between two people.
Building a “Personal OS” in ChatGPT: self-coaching with structured self-knowledge
Hiten shows his “personal OS” project—an always-on self-coaching environment containing his traits, tests, and writing guidance. The goal is faster, more accurate self-reflection and actionable coaching without starting from scratch each time.
Using the Personal OS for real workplace emotions and scenarios
Hiten demonstrates how minimal prompts can still work well once the project is context-rich. A brief workplace scenario (“someone trying to take over my project”) yields interpretation, motivations, and a path forward that resonates with him.
Framework + context = scalable coaching (and why it’s similar to executive coaching)
Claire connects the dots: traditional coaching uses assessments, frameworks, and 360 inputs—expensive and time-consuming. Hiten’s approach recreates that pipeline with projects and artifacts, making structured reflection accessible and iterative.
Demo: Winning by Design sales framework — turning PDFs into a reusable sales coach
Hiten walks through a project built from Winning by Design PDFs to generate discovery scripts and sales assets. The key is loading the framework documents as project files so the model reliably produces questions and structure aligned with the methodology.
Deep research + framework application: improving outputs with product/company context
They test using ChatGPT’s deep research to gather context on Claire’s product, then feeding it back into the sales framework project. When outputs don’t match intent, they treat it as a prompt/instruction bug and quickly iterate—highlighting the low cost of retries.
Lightning round: Hiten’s AI stack, resisting premature automation, and “bare metal” prompting
Hiten shares his tool usage (3–6 hours/day), mostly ChatGPT + Claude, plus a private desktop tool. He argues that many teams race to automation too early—before they’ve nailed reliable prompts and eval-like consistency through manual iteration.
When ChatGPT is wrong: direct correction, incentives, and avoiding “bribe prompting”
Hiten explains his practical debugging style: bluntly state what’s incorrect, propose what might be wrong (often instructions), and request a fix. He cautions against bribing or gimmicks, framing it as an incentives problem—especially with memory-enabled systems.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome