The IMPACT Method: How to Delegate to AI Like a Leader

Here is something I hear from leaders all the time: "I know AI can help, but I do not know where to start."

That is not a knowledge problem. It is a framework problem. Most people approach AI the way they approach a new piece of software — they open it up, poke around, and hope something clicks. But AI is not software in the traditional sense. It is more like a new team member. And like any new team member, it performs brilliantly when you give it clear direction and poorly when you wing it.

That is why I built the IMPACT Method. It is a six-step process for delegating work to AI the same way a great leader delegates work to people: with clarity, context, and intentionality.

You do not need to code. You do not need a computer science degree. You need to lead.

I — Identify the Machine Work

Before you touch any AI tool, step back and look at your week. What are you doing that follows a pattern? What tasks drain your energy without requiring your unique judgment?

This is where the Machine Work vs. Meaning Work framework comes in. Grab your calendar and your task list from the past week. Flag every item that is repetitive, rule-based, or low-judgment. Common candidates:

Email drafting and responses that follow a pattern
Meeting summaries and follow-up action items
Research compilation and competitive analysis
Report generation from existing data
Content repurposing across formats
Calendar and scheduling coordination

Pick one. Not five. One. The most impactful AI adoption starts with a single, clear use case.

M — Map the Current Process

Now document exactly how you currently do this task. Every step, every decision point, every input and output.

This matters because AI cannot improve a process you have not articulated. When a leader tells me "I just do it — it is intuitive," that is a sign that the process has never been examined. And unexamined processes are usually full of unnecessary steps.

Write it out like you are training a new hire:

I receive the input (an email, a data set, a request)
I check these criteria to decide what to do
I produce this output using these sources
I deliver it in this format to this audience

Be specific. The more precise your process map, the better AI will execute it.

P — Prepare the Context

AI is only as good as the context you give it. This step is about gathering everything the AI will need to do the job well.

Reference materials. Style guides, brand voice documents, templates, examples of good output.
Constraints. Word counts, formatting requirements, tone expectations, things to avoid.
Examples. Past work products that represent "good." AI learns from examples faster than from abstract instructions.
Domain knowledge. Industry jargon, company-specific terms, organizational context that an outsider would not know.

Think of this as building the briefing packet you would give to a sharp consultant on their first day. The more complete the context, the less back-and-forth you will need.

A — Assemble the Prompt (or Workflow)

Now you build the actual instruction. For simpler tasks, this might be a well-crafted prompt. For more complex workflows, it might be a chain of prompts or an agent configuration.

A strong AI prompt has four components:

Role. Tell AI who it is. "You are an executive communication specialist" performs differently than "You are a helpful assistant."
Task. State exactly what you want. Be specific about the deliverable, not just the topic.
Context. Provide the background, constraints, and reference materials from the Prepare step.
Format. Define what the output looks like. Length, structure, tone, and any formatting requirements.

Here is the difference between a weak prompt and a strong one:

Weak: "Summarize this meeting."

Strong: "You are an executive assistant preparing a summary for the VP of Sales. Using the attached transcript, create a summary with three sections: Key Decisions (bullet points), Action Items (owner + deadline), and Open Questions. Keep it under 300 words. Use a direct, professional tone — no filler language."

Same task. Vastly different output.

C — Commence (Run It)

Execute the workflow. But here is the key: do not judge it by the first output.

Most people try AI once, get an imperfect result, and conclude that AI does not work for their use case. That is like interviewing one candidate and deciding that hiring does not work.

Run the workflow multiple times. Compare outputs. Note where it consistently performs well and where it falls short. The first run is data, not a verdict.

Pay attention to:

Where is the output strong? (Keep those patterns)
Where is it off-target? (Refine your instructions)
Where is it missing information? (Improve your context)
Where does it over-deliver or go off-script? (Tighten your constraints)

T — Tune and Optimize

This is the step that separates the leaders who get 10% value from AI from the leaders who get 10x value.

Based on what you learned in the Commence step, refine your process:

Adjust the prompt. Add more specificity where the AI went off-track. Remove instructions that led to over-engineering.
Improve the context. Add examples of the output you want. Include examples of what you do not want.
Simplify the workflow. If a step is redundant or adds no value, cut it.
Set a quality bar. Define what "good enough" looks like. Not every AI output needs to be perfect — it needs to be a better starting point than a blank page.

Then run it again. And again. Most workflows hit their stride after three to five iterations. Once you have a tuned workflow that consistently produces good output, you have created leverage — that task now takes a fraction of the time it used to, every single time.

Putting IMPACT Into Practice

Let me walk through a real example. Say you spend 45 minutes after every client meeting writing a summary email.

Identify: Meeting follow-up emails. Classic Machine Work.

Map: You review your notes, pull out key decisions, list action items, add a friendly opening and closing, and send it within 2 hours of the meeting.

Prepare: You gather three examples of past follow-up emails you were happy with, your standard email signature, and a note about your preferred tone (warm but professional, no corporate jargon).

Assemble: You create a prompt that includes the role (executive relationship manager), task (write the follow-up email), context (meeting notes + examples), and format (under 250 words, specific structure).

Commence: You run it on your next meeting. The first version is 80% there — good structure, but the tone is slightly too formal.

Tune: You add "Match the casual-professional tone of the examples provided" to the prompt. The next version nails it.

What took 45 minutes now takes 5 minutes of review and a quick send. Over a year, that is roughly 150 hours reclaimed — almost a full month of working days.

The question is not whether that is worth doing. The question is: what will you do with those 150 hours?

That is where Meaning Work begins. That is where leadership, presence, and purpose get their time back.

You do not need to code. You need to lead. The IMPACT Method is how you start.