← Article 1 |

AI BUILDER'S PLAYBOOK • ARTICLE 2 OF 6

How to Talk to AI (So It Actually Helps)

February 6, 2026 • 12 min read

In Article 1, we established the foundation: AI is a smart intern, not a mind reader. You learned the three ingredients — Context, Task, Constraints. That's enough to get started. But "getting started" and "getting results you'd actually use" are two different things.

This article is about closing that gap. We're going to upgrade the three-ingredient framework into a five-part formula, then give you five business templates you can literally copy, paste, and fill in today. No theory. No jargon. Just the stuff that works.

By the end, you'll have a repeatable system for getting useful output from any AI tool — and you'll understand the three mistakes that sabotage most people's results before they even start.

The Five-Part Prompt Formula

Last time we covered Context, Task, and Constraints. Those three are the backbone. But there are two more ingredients that take your prompts from "decent first draft" to "I can actually use this right now." Here's the full formula:

R

Role: Tell it who to be

Give the AI a persona. This shapes everything about how it responds — the vocabulary it uses, the depth of its answers, and the assumptions it makes.

"You are an experienced operations manager who specializes in small service businesses."

C

Context: Tell it your situation

The background information the AI needs to give you something relevant. Who you are, what your business does, what's going on right now.

"I run a 12-person HVAC company in Atlanta. We do residential installs and service calls. Summer is our busiest season and we're about to hit it."

T

Task: Tell it what to produce

Be specific about the output you want. Not "help me with hiring" but exactly what deliverable you need.

"Write a job posting for a residential HVAC technician with at least 3 years of experience."

F

Format: Tell it how to structure the output

This is the upgrade most people miss. Tell the AI what shape you want the answer in. A bulleted list? A table? An email? A one-paragraph summary? This one change eliminates 80% of "that's not what I wanted" moments.

"Format it as a job listing I can post directly to Indeed. Include sections for: company overview, responsibilities, requirements, and benefits. Keep it under 400 words."

C

Constraints: Tell it what to avoid

The guardrails. What tone to use. What to leave out. What audience it's for. What length. This is where generic becomes specific.

"Don't use corporate jargon — we're a family-run shop and we want it to feel that way. Don't require a college degree. Do emphasize that we offer year-round work and paid training."

Role + Context + Task + Format + Constraints. Five parts. You don't need all five every time, but the more you include, the less fixing you'll do after.

Think of it like ordering at a restaurant. You can say "bring me food" and see what happens. Or you can say "I'd like the salmon, medium, with the vegetables instead of rice, no sauce on the side." Same kitchen. Wildly different results. The formula is just a way of being specific without having to think about what to be specific about.

Before and After: The Formula in Action

Let's see the difference when you stack all five parts together versus what most people actually type.

WRITING A PROPOSAL

WITHOUT THE FORMULA

"Write me a proposal for a landscaping job"

You get: A generic template that says things like "we look forward to the opportunity to serve you" and has blank placeholders everywhere. You spend 30 minutes rewriting it.

WITH THE FORMULA

"You are a sales writer for a landscaping company. I run Green Thumb Landscaping — we do residential design-build projects in the $15K-$50K range. Write a project proposal for a backyard renovation: new paver patio (400 sq ft), native plant garden beds, and landscape lighting. Format it as a professional PDF-ready proposal with sections for scope, timeline, investment, and terms. Keep the tone confident but warm. Don't include hourly rates — we price by project."

You get: A polished proposal you can actually send. Real sections, professional language, and your specific project details woven throughout. Five minutes of tweaking, not thirty.

Five Templates You Can Use Today

Here are five prompts for situations that come up every week in most businesses. Each one follows the formula. Copy them. Fill in the brackets. Paste them into any AI tool. They work in ChatGPT, Claude, Gemini — any of them.

1

Meeting Notes to Action Items

For when you walk out of a meeting with a page of messy notes and need to turn it into something your team can act on.

You are an executive assistant who specializes in meeting follow-ups.

Here are my raw notes from a [type of meeting] with [who was there]:

[Paste your notes here — messy is fine]

Turn these into a clean summary with three sections:
1. Key decisions made
2. Action items (who owns each one, deadline)
3. Open questions that still need answers

Keep it under one page. Use bullet points. Write it so someone who wasn't in the meeting can understand what happened.

Works great with: team meetings, client calls, vendor negotiations, board discussions

2

Customer Email to Professional Response

For when a customer sends you something and you need to respond quickly without sounding rushed or robotic.

You are a customer service manager for a [type of business].

A customer sent this email:
[Paste their email here]

Write a professional response that:
- Acknowledges their concern
- Explains [what happened / what we can do]
- Offers [specific resolution: refund, reschedule, discount, etc.]
- Keeps the door open for future business

Tone: warm and professional, not stiff or overly apologetic. Keep it under 150 words. Sign off as [your name/title].

Works great with: complaints, scheduling issues, pricing questions, warranty requests

3

Raw Data to Executive Summary

For when you have a spreadsheet or a pile of numbers and need to tell a story with them — for a partner, investor, or your own planning.

You are a business analyst helping a [type of business] owner understand their numbers.

Here's my data from [time period]:
[Paste your data — can be a table, CSV, or even a list of numbers with labels]

Write an executive summary that covers:
1. Top-line performance (revenue, volume, or whatever the main metric is)
2. What's trending up and what's trending down
3. The one or two things I should pay attention to
4. One specific recommendation based on the data

Format: 3–5 short paragraphs. No jargon. Write it like you're explaining to a smart business owner who doesn't have a finance degree. Include the key numbers but make them mean something.

Works great with: monthly financials, sales reports, website analytics, inventory data

4

Competitor Research to SWOT Draft

For when you need to think clearly about where you stand in your market, without hiring a consultant to tell you what you already half-know.

You are a strategy consultant helping a small business owner think through their competitive position.

My business: [what you do, your market, your size, what makes you different]

My main competitors:
- [Competitor 1: name, what they do, what they're known for]
- [Competitor 2: same]
- [Competitor 3: same]

Create a SWOT analysis for my business. For each section (Strengths, Weaknesses, Opportunities, Threats), give me 3–4 specific points — not generic business-school stuff, but things that are actually relevant to my market.

After the SWOT, add a section called "So What" with 2–3 strategic moves I should consider in the next 90 days. Be direct. If something's a real weakness, say so.

Works great with: annual planning, new market entry, pricing reviews, investor prep

5

Bullet Points to SOW/Proposal Draft

For when you've agreed on a deal verbally and need to put something professional on paper fast, before the momentum dies.

You are a professional proposal writer for a [type of business/consultancy].

I need a Statement of Work based on these bullet points from my conversation with the client:

[Paste your notes: what they want, timeline discussed, budget range, any specifics you agreed on]

Write a professional SOW with these sections:
1. Project Overview (2–3 sentences)
2. Scope of Work (numbered deliverables)
3. Timeline & Milestones
4. Investment (payment schedule)
5. What's Not Included (scope boundaries)
6. Terms (basic: payment net-30, one revision round, etc.)

Tone: professional but not stuffy. This is for a [small business / enterprise / nonprofit] client. Keep it under 2 pages.

Works great with: consulting projects, construction bids, design projects, retainer agreements

Notice the pattern in every template: you're filling in your specific details, but the structure does the thinking for you. That's the point. You shouldn't have to figure out how to ask — you should just have to fill in the blanks with what you already know about your business.

The Four Mistakes That Waste Your Time

Even with a good formula, there are patterns that consistently lead to frustrating results. Here's what to watch for.

1. Being vague about what "good" looks like

"Write me something professional" could mean a legal brief or a LinkedIn post. If you can't picture the output you want, the AI definitely can't either.

Fix: Before you type anything, ask yourself: "If someone handed me the perfect version of this, what would it look like?" Then describe that.

2. Skipping context because "it should know"

The AI doesn't know your industry's unwritten rules. It doesn't know that "competitive pricing" means something different for a plumber in rural Georgia than for a SaaS company in San Francisco. Every conversation starts fresh.

Fix: Spend 30 seconds on context up front. It saves 5 minutes of fixing later. Include your industry, size, location, and audience.

3. Cramming five tasks into one prompt

"Write my marketing email, update my pricing page, draft social posts for the week, and create a customer survey" — that's four different jobs. You wouldn't hand one employee four unrelated projects at the same time and expect quality on all of them.

Fix: One prompt, one task. You can do four prompts in a row and it'll take less time than fixing one bad mega-response.

4. Not iterating

This is still the biggest one. If the first response is 70% right, you're one sentence away from 95%. "Good, but make it shorter and less formal" takes five seconds to type and completely transforms the output.

Fix: Think of every first response as a draft. Your job is to direct, not to start over.

The "Show Your Work" Trick

Here's something that will immediately make you better at catching AI mistakes: ask it to show its reasoning.

AI tools are confident. They'll give you an answer that sounds great even when it's wrong. And because the output reads so smoothly, it's easy to miss errors — especially in numbers, dates, or recommendations.

The fix is dead simple. Add one sentence to the end of your prompt:

THE MAGIC SENTENCE

"Explain your reasoning step by step."

That's it. Six words that change everything.

When you ask AI to explain its thinking, two things happen. First, the AI actually produces better answers — the process of explaining forces more careful reasoning. Second, you can see where it went wrong if something doesn't add up. Instead of staring at a final answer wondering if you should trust it, you can trace the logic and spot the broken step.

WITHOUT "SHOW YOUR WORK"

"Should I lease or buy my next work truck?"

You get: "Leasing is generally better for businesses because of tax deductions." Sounds right, but is it right for your situation? No way to tell.

WITH "SHOW YOUR WORK"

"I drive about 35K miles a year for my plumbing business. Should I lease or buy my next work truck? Walk me through the financial comparison step by step."

You get: A breakdown showing that at 35K miles/year, most leases would hit mileage penalties — so buying makes more sense for you specifically. The reasoning is visible. You can check it.

This trick is especially valuable for anything involving numbers, comparisons, or recommendations. Don't just ask for the answer. Ask for the homework. Then check the homework. It takes 10 extra seconds of reading and it saves you from acting on bad advice.

When to Break It Into Steps

Some things are too big for a single prompt — even a well-structured one. A full business plan. A 20-page proposal. An entire marketing strategy. If you try to get everything in one shot, the output will be broad but shallow. Lots of words, not much substance.

The solution is prompt chaining: breaking a big task into steps and feeding the output of each step into the next one. Think of it like building a house. You don't tell the contractor "build me a house" and come back in six months. You do the foundation, then framing, then plumbing, then electrical, one stage at a time.

EXAMPLE: BUILDING A MARKETING PLAN IN FOUR PROMPTS

1

Identify your audience

"I run a mobile pet grooming service in Austin. Help me identify my top 3 customer segments — who they are, what they care about, and where they hang out online."

2

Pick your channels

"Based on those three segments, which marketing channels should I focus on? I have a budget of $500/month and about 5 hours a week. Rank them by expected impact."

3

Create the content plan

"For the top 2 channels, create a 4-week content calendar. Include post topics, frequency, and any specific calls to action. I want to book 10 new recurring clients this quarter."

4

Write the first batch

"Write the first week's posts from that calendar. Match my brand voice — casual, friendly, a little funny. Include one before/after photo prompt for each post."

Each step takes 30 seconds to prompt and builds on the last. In 10 minutes you have a real marketing plan with actual content — not a vague strategy document.

The key insight: you stay in control at every step. After step one, you can say "actually, remove segment three, they're not really my customer." After step two, you can say "I don't want to do TikTok, swap that for email marketing." Each checkpoint keeps the output honest and aligned with what you actually know about your business.

A good rule of thumb: if your prompt has the word "and" more than twice, you should probably break it up. Each prompt should ask for one clear deliverable. Chain them together and you get something much better than trying to do it all at once.

Your Cheat Sheet

Print this. Stick it next to your monitor. Refer to it every time you open an AI tool until it becomes automatic.

THE FIVE-PART FORMULA

R

Role — Who should the AI be?

"You are a [job title] who specializes in [area]."

C

Context — What's my situation?

Business type, size, industry, what's happening right now.

T

Task — What specific output do I need?

Not "help me with X." What deliverable am I asking for?

F

Format — What shape should the output be?

Email? Bullet list? Table? How long? What sections?

C

Constraints — What should it avoid or include?

Tone, audience, things to skip, things to emphasize.

Bonus: Add "Explain your reasoning step by step" to any prompt where accuracy matters.

AI BUILDER'S PLAYBOOK

2 How to Talk to AI (So It Actually Helps) You're here
4 The Iteration Loop: Getting AI to Improve Its Own Work Coming soon
5 When AI Gets It Wrong (And What to Do About It) Coming soon
6 From One Tool to a System: Scaling AI in Your Business Coming soon

What's Next

You now have a formula and five templates. That's enough to start using AI productively every single day. But knowing how to talk to AI is only half the equation. The other half is actually building something with it.

Next in the series: Your First Build — From Idea to Working Tool — where we walk through building a real business tool, step by step. No coding required. You'll have something you can use this week.

Want help putting this into practice?

Textstone Labs helps business owners turn AI from a curiosity into a daily tool. Book a free 30-minute call and we'll build custom prompts for your three most time-consuming tasks.

Let's Talk →

Enjoying the series?

Get the next article in your inbox. No spam — just practical AI guides for business owners.

Textstone Labs — AI implementation for people who build things.