Alicia Lyttle: Stop Using AI Like Google. Build a Team.

We have talked a lot about leverage throughout this Women & Wealth series.

Ariel Pryor talked about escaping the ceiling of selling only her time.

Anne Mahlum talked about building systems that allow a company to scale beyond the founder.

Then Alicia Lyttle walked onto the stage and introduced another form of leverage.

Artificial intelligence.

But her argument was not simply that we should use ChatGPT.

At this point, almost everyone in the room already did.

Her question was much more interesting.

That distinction became the foundation of her presentation.

Most people ask AI a question.

• Write this email.
• Summarize this document.
• Give me some ideas.
• Rewrite this caption.

Alicia wants us thinking more like organizational designers.

• What work am I repeatedly doing?
• What requires my judgment?
• What requires my relationships?
• What actually creates revenue?
• What can be delegated?

And if some of that delegated work no longer requires another human being, what would an AI team around me actually look like?

That was where the presentation became much more useful than another list of ChatGPT prompts.

Alicia Has an AI Employee Named Maximus

Alicia opened with a problem most people understand immediately.

Email.

She said she had not manually managed her email inbox for 165 days.

Instead, she had built an AI agent named Maximus.

According to the numbers she presented, Maximus had also eliminated thousands of promotional messages, performed hundreds of inbox scans, sent emails, and created graphics.

Whether every business should allow an AI agent that much autonomy is a separate question.

What interested me was the way she thought about the problem.

She did not ask how AI could help her answer email faster.

She asked why she was doing it at all.

That is a much more valuable question.

Entrepreneurs often use technology to become faster at low-value work instead of asking whether they should remain responsible for that work in the first place.

• We automate a bad process.
• We create templates for repetitive work.
• We become extremely efficient at things the founder should probably have stopped doing years ago.

AI gives us an opportunity to reconsider the workflow itself.

Her Org Chart Was the Real Presentation

Alicia showed an organizational chart from 2025.

Some names represented humans.

Many represented AI agents.

Then she showed how that model was evolving.

Instead of one founder creating all the AI agents, each human team member could eventually have her own AI team underneath her.

That got my attention.

Imagine a marketing manager who has:

• An AI research assistant
• An AI copy assistant
• An AI analytics assistant
• An AI content repurposing assistant
• An AI meeting and project coordinator

The human does not disappear.

Her capacity changes.

One capable person may now be able to manage an amount of output that previously required several more people.

That creates a completely different organizational question.

That does not automatically mean fewer humans.

In some companies, increased productivity will allow a smaller team to produce the same output.

In others, the same team may produce dramatically more.

And in growing businesses, AI may allow the company to delay certain hires while revenue catches up.

The important point is that the org chart itself is becoming flexible.

Start With the Work, Not the Technology

I liked Alicia's process for identifying an AI team because she began with something practical.

• What are you doing repeatedly?
• What takes too much time?
• What is administrative?
• What do you dislike doing?
• What creates bottlenecks?

Then define a role around that work.

1. 01Give the AI agent a job.
2. 02Give it responsibilities.
3. 03Give it instructions.
4. 04Give it a name if that helps make the role easier to conceptualize.
5. 05Review what it produces and improve the process.

This is important because people often approach AI backwards.

They discover a new tool and ask what they can do with it.

A better question is what problem in the business needs solving.

Then determine whether AI is actually the right solution.

Technology should follow the workflow. Not the other way around.

Her First Prompting Lesson Was Also Probably Her Best

Before asking AI to create something, let it interview you.

That sounds simple.

It can dramatically change the quality of the result.

Instead of asking it to write your sales page, try something closer to this:

Now AI is not filling every missing detail with assumptions.

It is gathering context.

This is especially useful because weak AI outputs are often not really AI failures.

They are context failures.

We give three sentences of information and expect the system to somehow understand:

• Our customer.
• Brand.
• Strategy.
• History.
• Pricing.
• Voice.
• Constraints.
• Competitive position.
• Objectives.

Then we complain that the answer sounds generic.

Alicia's framework flips the responsibility.

If the system does not know enough, make it ask.

That is something I will use.

Tell AI What Good Looks Like

Alicia then went through several techniques for improving output.

One was giving the work a high standard.

The important part here is not the dollar amount.

AI does not suddenly unlock a secret $10,000-an-hour brain because we typed a large number.

The useful part is setting an explicit quality bar.

What does excellent mean?

• Sophisticated?
• Concise?
• Commercially aware?
• Persuasive without sounding desperate?
• Written for a CEO?
• Written for someone skeptical?
• Written by someone with twenty years of domain expertise?

The more clearly we define the standard, the easier it becomes for the model to attempt to meet it.

Give It a Specific Perspective

Alicia made the same point when assigning roles.

Again, I would not interpret this as AI magically becoming that person.

It is prompting.

But good prompting constrains the problem.

And constraints can improve output.

This is exactly what we do with humans.

If I hired a consultant and said "help with my business," that is an enormous mandate.

If I said "analyze why qualified customers are reaching the final stage of our sales process but not closing, and identify the three highest-probability interventions we should test first," I have made the assignment dramatically better.

AI benefits from the same clarity.

Show Examples Instead of Endlessly Describing Your Taste

This may have been Alicia's most immediately useful content lesson.

If you already have something that works, give it to the AI.

• Your strongest email.
• Your best-performing advertisement.
• The proposal that closes.
• The caption that generated unusual engagement.
• The sales call transcript that converted.
• The customer testimonial that perfectly communicates the value.

Then ask:

That is much better than repeatedly typing "make it sound more like me."

If you want the AI to understand your taste, show your taste.

This is true far beyond writing.

Examples create a target.

Never Trust the First Answer

Alicia repeatedly told the audience not to accept the first AI output.

I strongly agree.

Her approach was to force the model to challenge itself.

Ask for multiple perspectives.

• What would the salesperson think?
• What would the skeptical customer think?
• What would the marketer think?
• What would a devil's advocate say?

Then compare them.

She also suggested asking AI to critique its own work.

1. 01Ask what the strongest objections someone could raise against this proposal are.
2. 02Then ask it to rewrite the proposal to address the legitimate objections it identified.

That turns AI from a generation tool into a stress-testing tool.

And I think that is one of the most underused ways to work with it.

Make Different Models Disagree

Alicia took the stress-testing idea one step further.

1. 01Create something in one AI system.
2. 02Take it to another.
3. 03Tell the second system to tear it apart.
4. 04Return the criticism to the first and ask it to rebuild.

The principle is useful even without switching platforms.

Do not use AI only to agree with you.

This matters enormously.

AI can become a very sophisticated yes-man if that is how we prompt it.

I think this idea is especially valuable for entrepreneurs because founders already suffer from confirmation bias.

We become attached to:

• Our product.
• Our pricing.
• Our strategy.
• Our hire.
• Our acquisition.
• Our brilliant new idea.

AI becomes more valuable when we ask it:

• Why might I be wrong?
• What am I overlooking?
• What would make this fail?
• What evidence would change this recommendation?
• What is the strongest argument for doing the opposite?

That is closer to strategic thinking than simply asking AI to validate what we already want to do.

About Genius Mode

Alicia demonstrated something she called genius mode.

She described telling ChatGPT to operate as an elite strategic thinker with higher standards for reasoning, creativity, and clarity.

She also talked about creating other modes.

• Customer mode.
• Lawyer mode.
• Sales strategist mode.
• Skeptical buyer mode.

Later in the presentation, she made an important clarification.

These modes are made up.

That is exactly how I would understand them.

There is not a magical hidden genius mode that transforms an AI system into a fundamentally smarter model.

The mode is a set of instructions.

But that does not make the technique useless.

Quite the opposite.

She is essentially creating reusable thinking frameworks.

Instead of repeatedly explaining that you want it to challenge your assumptions, identify second-order effects, think commercially, separate fact from inference, examine downside risk, and propose alternatives, you can create a named framework containing those instructions.

Then reuse it.

The power is not in the name. The power is in the quality of the instructions behind the name.

Personalization Compounds Usefulness

Alicia repeatedly used another phrase throughout the presentation.

Her point was that an AI system becomes much more useful when it has sufficient context about your goals, business, customers, skills, experience, resources, constraints, writing style, and prior decisions.

The more relevant context available, the less time we spend starting every conversation from zero.

I agree.

But this deserves a security qualification.

The Million-Dollar Prompt Was Not Really About the Million Dollars

Alicia gave entrepreneurs a long prompt asking AI to create a plan for generating $1 million over the next twelve months.

She told the audience to have AI analyze:

• Skills, experience, and network
• Business, resources, and goals
• Customer problem and offer
• Pricing and sales strategy
• Marketing channels and partnerships
• Monthly targets

Then create a 90-day plan and immediate actions.

The million-dollar headline is obviously exciting.

But the underlying mechanism is more valuable than the number.

She is asking AI to function as a persistent strategic planning partner.

1. 01Day one: what should I do?
2. 02Report back: here is what happened. What changes?
3. 03Day two: what should I do now?

That feedback loop is much more interesting than receiving one beautiful 30-page business plan and never looking at it again.

Strategy should update when reality arrives.

AI makes continuous planning much easier.

AI Can Recommend. It Cannot Guarantee.

This is where I would add an important limitation.

An AI-generated plan for making $1 million is not evidence that the market will produce $1 million.

• It does not create demand.
• It cannot guarantee customers.
• It may misunderstand the market.
• It may make assumptions that sound intelligent but are wrong.
• It may overestimate conversion.
• Underestimate costs.
• Invent facts.
• Recommend a strategy that worked historically but no longer works now.

The human still has to validate the plan against reality.

That is why the word I prefer is copilot, not oracle.

Use AI to accelerate thinking.

Then test the thinking.

The Networking Prompt Showed What Good AI Use Actually Looks Like

Alicia gave another example that initially sounded ordinary.

Find networking events near me.

But then she layered intelligence onto the task.

1. 01Understand my customer.
2. 02Understand my business goals.
3. 03Find relevant events geographically.
4. 04Rank them based on likely return on time.
5. 05Tell me which five matter most.
6. 06Prepare a short introduction appropriate for those rooms.
7. 07Give me conversation starters.
8. 08Give me the follow-up strategy.

That is the difference between search and workflow.

A search engine gives information.

A well-designed AI workflow can potentially help transform information into a sequence of actions.

That is where a lot of value will come from.

Build an AI Team Around What You Should No Longer Be Doing

Alicia eventually brought the presentation back to the org chart.

Her exercise was straightforward.

1. 01Tell AI about your business and daily workflow.
2. 02Ask it to identify three to five repetitive or time-consuming responsibilities you currently perform but probably should not.
3. 03Create an AI role around each responsibility.
4. 04Give each one a title, responsibilities, boundaries, instructions, expected outputs, and a way to escalate problems.
5. 05Decide how the human interacts with it.

This is where I think entrepreneurs should spend serious time.

That is a much better question.

AI Delegation Should Resemble Human Delegation

There is another principle I would add to Alicia's framework.

We should delegate to AI with many of the same disciplines required when delegating to people.

A bad manager tells an employee to handle marketing, then becomes angry when the result is not what she imagined.

A better manager defines:

• Objective
• Context
• Authority
• Constraints
• Deadline
• Quality standard
• Examples
• Metrics
• Escalation rules

The same applies to AI agents.

Now we have a process.

AI does not eliminate management.

It makes good management more important because execution can happen much faster.

The Human Should Move Upward

Near the end, Alicia's AI-team exercise produced a simple recommendation for her own role.

She should primarily be:

• Thinking.
• Selling.
• Teaching.
• Building relationships.

Everything else should increasingly be delegated.

I would not use those exact four categories for every founder.

But I agree with the direction.

The founder should continually move toward the activities where her judgment, credibility, relationships, creativity, decision-making, and capital allocation produce disproportionate value.

That is exactly what Anne Mahlum said about earning the CEO title every six months.

The difference is that Alicia is showing another mechanism for doing it.

Previously, scaling required hiring people beneath us.

Now the answer may increasingly be:

• Hire.
• Automate.
• Build software.
• Deploy an AI agent.
• Or combine all four.

The AI Advantage May Really Be an Organizational Advantage

It is tempting to think the winners of the AI era will simply be the people who know the best prompts.

I doubt that.

• Prompts will become easier.
• Models will improve.
• Interfaces will improve.

Many techniques that feel advanced today will eventually become normal features.

The larger advantage may belong to people who learn how to redesign organizations around the technology.

• Who does what?
• What gets automated?
• Where does judgment remain human?
• What information moves where?
• What requires approval?
• How many customers can one employee now support?
• How quickly can the company test an idea?
• How much administrative work can disappear?
• How much more time can be redirected toward customers and revenue?

Those are business questions. Not AI questions.

And that may have been Alicia Lyttle's most important contribution to Women & Wealth.

She was not really teaching us how to talk to a chatbot.

She was asking us to reconsider the shape of work.

This Is Part Seven

This is Part Seven of my Women & Wealth conference series.

What I am enjoying about working through these presentations is how the ideas are beginning to connect.

Ariel showed how products create leverage beyond time.

Anne showed how systems and delegation create organizations that scale beyond the founder.

Alicia introduced the next layer.

I think every business owner should leave her presentation with one exercise.

Write down everything you did last week.

Then classify it.

1. 01What required you?
2. 02What required a competent human?
3. 03What could software automate?
4. 04What could AI assist with?
5. 05What could an AI agent eventually own with appropriate controls?
6. 06And what should simply stop being done?

Because the goal is not to have the most AI.

The goal is to build a business where human attention is spent on the things for which human attention is actually valuable.

That is a much more interesting definition of productivity.

And potentially, a much more powerful definition of leverage.