In this conversation, Cary Weston, host of The ChatGPT Experiment, reveals a strategic approach to leveraging ChatGPT for complex, repetitive research tasks, moving beyond superficial interactions to a more collaborative problem-solving model. The core thesis is that by treating ChatGPT as an "amazing intern" and engaging it in a structured, multi-stage dialogue, users can delegate the "busy middle" of research and analysis, freeing up valuable time for higher-level ideation and refinement. This method uncovers the hidden consequence of rushed prompts: dissatisfaction and wasted effort. By investing time upfront in defining the "what," "why," and "what success looks like," and then inviting ChatGPT to ask clarifying questions, users can unlock its potential as a thinking partner. This approach is crucial for anyone seeking to automate processes, reduce time spent on manual tasks, and gain a competitive advantage by efficiently extracting actionable insights from vast amounts of information. It offers a clear path for both experienced and novice ChatGPT users to achieve more consistent and valuable outcomes.

The Hidden Power of Dialogue: Unlocking Scalable Research with ChatGPT

The immediate impulse when encountering a complex, time-consuming task is often to find the quickest solution. We might think, "How can I just get this done?" This natural inclination, however, can lead us down a path of superficial fixes that fail to address the underlying systemic issues. In a conversation on The ChatGPT Experiment, host Cary Weston demonstrates a counterintuitive approach to using AI tools like ChatGPT, revealing how the most effective solutions often emerge not from demanding immediate answers, but from a deliberate, structured dialogue. Weston argues that many users rush the interaction, leading to inconsistent results and a perception that the tool is unreliable. The real value, he explains, lies in treating ChatGPT not as a simple command-line interface, but as a collaborative partner capable of deep problem-solving. This requires a shift in perspective: moving from a transactional request to a more nuanced, iterative conversation that maps out consequences and builds a robust, scalable solution.

Unpacking the System: From Manual Drudgery to AI Collaboration

In this episode, Cary Weston walks through a real-world coaching scenario where he guided a trade advocacy group through the process of automating their time-consuming research on legislative and regulatory issues. The group, facing the challenge of tracking changes across 50 states, had previously relied on manual searches of various websites and publications, a process that consumed significant time and mental bandwidth. Their attempts with ChatGPT had yielded inconsistent results, leading them to revert to their manual methods. Weston's approach, however, was designed to circumvent the common pitfalls of AI interaction, focusing on a methodical, consequence-aware strategy.

The "Busy Middle": Where Time and Energy Vanish

Weston introduces a core concept in his work: the "busy middle." He posits that most projects, whether AI-assisted or not, can be divided into three phases: ideation/planning, the busy middle, and editing/perfecting. He observes that individuals and teams often spend an inordinate amount of time and energy in the busy middle--the execution phase where data is gathered, filtered, and processed--while neglecting the crucial upfront planning and the final refinement stages. This is precisely where ChatGPT can become an invaluable ally, but only if approached correctly.

"I would argue that most of us spend far too much time in the busy middle and that's where our time and our mental bandwidth gets exhausted and far too little time in the editing and perfecting phase," Weston explains. "Our value is really in that third phase... and in the first and the third phase... what are we doing in planning and then the ideation perfecting that's where our experience and our value gets put in but our time gets sucked into that busy middle."

The advocacy group's problem perfectly exemplified this. Their manual research involved sifting through numerous state resources, industry websites, magazines, and news outlets--a classic "busy middle" scenario. They were manually filtering information, re-explaining details, and visiting inconsistent sources, all of which contributed to inefficiency and potential oversight.

The Four-Part Framework: Building a Foundation for Collaboration

Weston's strategy hinges on a structured four-part framework for communicating with ChatGPT, treating it as an "amazing intern." This framework is not about simply issuing commands, but about establishing a clear context and fostering a collaborative environment:

1. What are we doing? Clearly define the task or objective.
2. Why are we doing it? Provide the underlying purpose and motivation. This is critical for enabling ChatGPT to understand the broader implications and context, going beyond the literal task.
3. What does success look like? Articulate the desired outcome and the criteria for achieving it.
4. Do you have questions for me? This step is crucial for opening a dialogue, allowing ChatGPT to identify gaps in information and ensure it has everything it needs for optimal performance.

This framework, Weston emphasizes, triggers ChatGPT's reasoning and logic components, enabling it to think more deeply about the task and its objectives.

From Transcript to Project: Designing for Scalability and Flexibility

For the advocacy group, Weston began by gathering information through dialogue. He engaged them in a conversation, using his four-part framework to understand their current process, the reasons behind it, and their vision of success. He then assigned them homework: to provide specific URLs of the resources they regularly checked and to describe what success would look like if ChatGPT could automate parts of their task.

The collected information--the transcript of the conversation and the follow-up email with URLs and success criteria--became the raw material for a conversation with ChatGPT itself. Weston’s initial prompt to ChatGPT was not to solve the problem directly, but to act as a strategic advisor in building a step-by-step plan for a repetitive goal.

"I have a big task for you and I need you to serve as my strategic advisor to help me build out a step by step plan we can develop to achieve a repetitive goal," Weston stated. He then provided the transcript and the homework results, instructing ChatGPT to review them and engage in a dialogue to map out the best path forward. This approach deliberately skipped the step of asking ChatGPT to "fix the problem" or "give me the output." Instead, it invited ChatGPT to be a problem-solving partner, thinking alongside Weston.

ChatGPT responded by accurately identifying the core issues: inconsistent sources, repeated research, manual filtering, and re-explaining information. It then proposed a path forward, suggesting the creation of an AI assistant with a defined job description, clear output expectations, defined search parameters, and reporting frequency. Critically, it asked a clarifying question: "if you had to choose one primary outcome, do you want a recurring summarized tool or do you want a deeper analysis?" This dialogue allowed Weston to refine the objective--a recurring summarized tool--and guided the subsequent steps.

The Modularity Advantage: Building a "Project" with a Brain and Resources

Weston’s ultimate goal was to create a ChatGPT "Project"--a feature designed for persistent, purpose-driven tasks and capable of remembering past interactions. He explained that unlike a custom GPT, which acts as a stateless tool, a project has memory, allowing it to build upon itself over time.

A key insight here is the importance of modularity. Weston recognized that the underlying research landscape would change--sources might be updated, issues might evolve, and criteria for what constitutes a "material change" could shift. To avoid creating a rigid system that would quickly become obsolete, he opted for a modular design.

"I don't want to laminate anything... because the minute you laminate something you can't change it," Weston stated. "So I want to have something that's flexible. I want to have something that can be changed."

This led to the concept of a "brain" (the core instructions for the project) and separate, modular "resources" (supporting documents). The brain would contain the overarching instructions--what the task is, why it's being done, and the general goals. The resources would contain the specific, changeable details.

The project was structured with four key resources:

1. Sites to Monitor: A document listing the trusted URLs that ChatGPT is permitted to access. If a source is not listed, it must be ignored. This document can be updated with new URLs or by removing outdated ones without altering the project's core instructions.
2. Material Change Criteria: A document defining what constitutes a significant change in legislation or regulation, and what does not. This allows for granular filtering, ensuring that only relevant updates are flagged. This document can also be updated as the group’s needs evolve.
3. Time and Look Back Rules: A document specifying the frequency of reports (e.g., weekly) and how far back ChatGPT should look for changes. Weston noted that ChatGPT can be poor with dates, so this document includes instructions for determining the current date and then applying the look-back period.
4. Output Template: A document defining the exact structure and formatting of the final report. This ensures consistency and meets the "what success looks like" criteria. This template can be modified to change the output format without reconfiguring the entire project.

This modular approach ensures that the core "brain" of the project remains stable, while the "resources" can be easily updated. As Weston puts it, "As long as the brain knows what files you have, what they're there for, and how to use them, you can change those reference documents as you wish without changing the project as a whole." This significantly reduces the effort required for ongoing maintenance and adaptation.

The Competitive Edge: Patience and Upfront Investment

The entire process--from the initial dialogue with the group to the structured conversation with ChatGPT and the creation of modular resources--represents a significant upfront investment. This is precisely where the competitive advantage lies. Most individuals and teams, Weston observes, rush into demanding immediate outputs from AI, leading to frustration when results are inconsistent or incomplete.

"A little bit of work right, let ChatGPT do the heavy lifting, engage it in a dialogue, have it think with you, have it create resources for ChatGPT," Weston advises. "Have it do the busy middle, have it do the heavy lifting and utilize those resources like your call transcripts or emails or whatnot."

The advantage is gained by those willing to do the "hard work" upfront: clearly defining the problem, articulating the purpose, specifying success, and engaging the AI as a partner. This patient, systematic approach ensures that the AI is configured to perform the "busy middle" tasks effectively, freeing up human cognitive resources for higher-value activities like strategic thinking, analysis of the AI's output, and decision-making. The modular design further amplifies this advantage by ensuring the solution remains adaptable and durable over time, a quality often lacking in quick-fix approaches. By investing in this deeper understanding and structured implementation, users can transform ChatGPT from a novelty into a powerful, scalable engine for productivity and insight.

Key Action Items

• Embrace the Four-Part Framework: Before engaging ChatGPT for any significant task, clearly define: 1) What you are doing, 2) Why you are doing it, 3) What success looks like, and 4) Invite ChatGPT to ask clarifying questions. (Immediate Action)
• Delegate the "Busy Middle": Identify repetitive, time-consuming tasks in your workflow and use the four-part framework to structure a dialogue with ChatGPT, enabling it to perform the data gathering, filtering, and initial processing. (Immediate Action)
• Design for Modularity: When building ChatGPT Projects or complex prompts, separate core instructions ("the brain") from dynamic data or criteria ("resources"). This allows for easier updates and long-term adaptability. (Immediate Action)
• Prioritize Dialogue over Demands: Instead of asking ChatGPT to "do X," engage it as a strategic advisor or partner. Ask it to help you build a plan, map out steps, or identify potential issues. (Immediate Action)
• Invest in Upfront Clarity: Recognize that the time spent defining the problem, purpose, and success criteria upfront will significantly improve the quality and consistency of ChatGPT's output, preventing rework and frustration. (Immediate Action)
• Develop a "Project" Mindset: Explore ChatGPT's "Projects" feature for tasks requiring memory and iterative improvement. This offers a more persistent and evolving AI assistant compared to stateless custom GPTs. (Over the next quarter)
• Iterate and Refine: Treat your AI-assisted solutions as living systems. Plan to regularly review outputs, identify areas for improvement, and update modular resources to maintain effectiveness and adapt to changing needs. (Ongoing, with specific review cycles e.g., quarterly)