Delegating Tasks to AI Using a Four-Part Framework

Many professionals view AI as a creative engine, which is a mistake that causes unnecessary anxiety and poor results. By shifting your perspective from "AI as a creator" to "AI as a delegatable intern," you can bypass the fear of factual errors and focus on the high-leverage work of execution. This shift allows you to treat AI as a partner for repetitive, time-consuming tasks rather than a source of truth. Your competitive advantage comes not from the tools themselves, but from your ability to apply a rigorous, four-part delegation framework to your existing workflows. If you adopt this systemic approach to prompting and personalization, you will turn AI from a confusing, blinking cursor into a force multiplier that grows in value over time.

The execution-first paradigm

The most common barrier to AI adoption is the creation trap. When users ask AI to generate content from scratch, they often encounter factual errors or hallucinations, which creates a feedback loop of distrust. Cary Weston suggests that this anxiety is a byproduct of using the wrong mental model. By prioritizing execution over creation, you narrow the scope of the AI responsibility, making its output verifiable and actionable.

"I find the real value in AI executing something doing a task that I can explain to it, that I can design and give it to so that it can make my life easier."

-- Cary Weston

When you move from "create me a marketing plan" to "execute this specific formatting task on this existing data," you shift the system from one that requires constant supervision to one that provides immediate utility. This is the difference between a tool that creates more work for you and one that systematically clears your backlog.

The four-part delegation framework

Treating AI as a simple search tool leads to one-sentence, Google-style prompts that inevitably fail. Weston suggests that the system responds best when you treat it like a human intern. This requires a specific protocol, a four-part framework, that forces you to define parameters before the AI starts working.

"The four-part framework is what are we doing? Why are we doing it? What does success look like if we do it right? And do you have any questions from me?"

-- Cary Weston

This framework creates a structural buffer against ambiguity. By explicitly asking, "Do you have any questions for me?" you open a feedback loop that allows the AI to clarify its objectives. Over time, this recursive process, where you refine the prompt based on the AI answers, builds a more resilient and accurate workflow than a single, high-stakes prompt ever could.

Personalization as a systemic moat

Most users treat AI tools as static, out-of-the-box solutions, failing to realize that the system is designed to adapt to the user. Personalization, which means providing ongoing feedback on writing style, formatting preferences, and organizational rules, is the mechanism by which you create a lasting advantage.

When you consistently feed the system instructions on what to do more of and what to avoid, you are training a specialized assistant. This creates a compounding effect: the more you personalize the tool, the less prompt engineering you need to do in the future. While others struggle to get consistent results because they treat every session as a blank slate, your personalized system becomes more efficient and accurate with every interaction, creating a significant productivity gap over months and years.

Key action items

• Audit your daily tasks: Identify one repetitive or frustrating task that takes you too long. This is your primary candidate for AI delegation. (Immediate)
• Apply the four-part framework: Before your next interaction, write down: 1. The task, 2. The why, 3. The definition of success, and 4. A prompt for the AI to ask you clarifying questions. (Immediate)
• Shift to execution prompts: Stop asking the AI to create and start asking it to execute on data you provide, such as summarizing meeting transcripts or formatting existing notes. (Immediate)
• Create a personalization document: Spend 30 minutes documenting your preferred writing style, common formats, and rules you want the AI to follow. Input this as custom instructions or a system prompt. (Over the next quarter)
• Establish a feedback loop: After every major task, tell the AI what it did well and what it missed. This habit creates a compounding improvement in output quality that pays off in 12 to 18 months. (Ongoing)