My Human-Centric Handbook for LLM Communication

How to sharpen the tool without losing your humanity.

In the rush to embrace Artificial Intelligence, most of the conversation has centered on speed, how to get answers faster, how to automate more, and how to "hack" the prompt box. But as we integrate Large Language Models (LLMs) into our professional and creative lives, we are discovering that velocity without a roadmap leads to lame, genetically modified results and questionable tactics (like my LinkedIn friend).

True mastery of an LLM isn't about knowing the right magic words. It’s about a fundamental shift in mindset. It requires treating the AI as a powerful but public consultant that needs a clear boundary, a sharp strategy, and, most importantly, a human director (that’s you!)

This is inspired by the clear evidence of countless AI-slop articles and blog posts I’ve seen recently, and by the hundreds or thousands of AI-generated books being uploaded to Amazon.

The Privacy Paradox: Communicating with Caution

The most dangerous mistake a user can make isn't a "bad prompt," it’s a lack of boundaries. Many treat the interface like a private vault, but in reality, every prompt is a contribution to a larger data cycle.

The Golden Rule of Input: Always strip Personally Identifiable Information (PII). This isn't just about your name; it’s about protecting your business logic, your clients' data, and your brand's proprietary interactions and perhaps your own personal health records.

The Trade-off: Convenience vs. Control

Unless you are using enterprise-grade, "zero-retention" APIs, your data is often collected for training, reviewed for safety, and can persist even after deletion. Treat the LLM as a brilliant but public consultant. Feed it the logic of your problem, not the identity of your subjects.

Sharpening the Axe: Accuracy Over Velocity

In an era of instant answers, the most effective communicators are actually the slowest. High-level output requires a "sharpening the axe" phase. As Abraham Lincoln famously noted, if he had six hours to chop down a tree, he’d spend the first four sharpening the axe.

• The Power of the Iterative Loop: Avoid "One-Shot" prompts. Work in a conversational back-and-forth to define parameters and refine logic.

• Quality over Speed: If you spend 70% of your time "sharpening" the context and the final 30% on the actual generation, the result will be significantly more accurate.

The Expert Persona: Beyond Generic Roles

Defining a persona is common, but the most effective communicators anchor the AI’s logic in the work of established experts.

• The "Mentor" Shortcut: Reference specific thinkers whose work is in the training data and public domain. Think of this as thought leaders that the AI agent can be inspired by as it works with you.

• Role-First Logic: Setting this role at the start provides a lens through which the AI will interpret all subsequent dialogue.

The Human Anchor: Protecting Original Thought

Perhaps the most valuable insight in the age of AI is knowing when to stop using it. The project must be completed by you. We were able to complete great work before AI was around, and that good ole human-generated intelligence is still the gold standard (or risk losing it forever).

• AI as Scaffolding, Not Architecture: Use the AI to organize or brainstorm, but the soul of the work, the spark and conviction, must be your baby.

• The Self-Reliance Rule: The AI is a luxury of efficiency, not a necessity for intelligence.

Directed Brainstorming: Concept-First Collaboration

High-value communication isn't about pulling ideas out of thin air; it’s about bringing a solid, human-generated concept to the table.

• The Spark vs. The Flame: You provide the "spark" (the core concept), and the AI helps grow it into a "flame" (the execution).

• Active Stewardship: You remain the director. If the AI drifts, you steer it back to your original vision. Once we stop directing, we become mindless consumers. Let’s not let that happen.

The Organic Finish: Proofreading vs. Processing

The final step is a rigorous human audit to ensure the content isn’t genetically modified, highly processed junk.

• The Intuition Audit: Read the output aloud to catch algorithmic phrasing. If it doesn't sound like you, it shouldn't be part of the final product.

• External Validation: Use a separate grammar tool for a final review to ensure the result is fresh, organic, and not fillled with tiepos.

• Final Ownership: The goal is a result that is technically sound, mechanically perfect, and above all, authentically human. Like, legit human!

The future of AI communication doesn't belong to those who type the fastest, but to those who think most clearly. By prioritizing privacy, embracing a slow and iterative sharpening process, and insisting on human-led creativity, we ensure that the tools we use enhance our work rather than dilute it. AI can assist the brainstorm, but the final result must always bear the unmistakable mark of human original thought. Keep your content organic, keep your data secure, and never let the machine take the wheel of your best ideas until we have no other choice, and it kills us all (hopefully not).

It’s local election time in Nashville (Davidson County), and each year I’m left scratching my head on who to vote for. Most people don’t have the time to research every candidate, so they don’t vote at all. But I have a solution for you, no matter where you are in the country, because this matters.

Check your voter status…

REGISTER TO VOTE / CHECK VOTER STATUS

Most of the services you interact with every single day are managed at the local level, not by the White House. When you vote for a Mayor, City Council member, or County Commissioner, you are choosing the people who decide:

• Public Safety: How your local police and fire departments are funded and trained.

• Infrastructure: Which potholes get filled (if any!), how traffic is managed, and where bike lanes or sidewalks are built.

• Education: School board members determine the curriculum, teacher pay (they need more), and safety protocols for your neighborhood schools.

• Zoning: Whether that empty lot across the street becomes a park, a high-rise, a grocery store, or another car wash. What’s up with all of these car washes?

Shockingly, our local/primary elections often have voter turnout of 10% to 30%. Then we all complain about the above. And before you scoff at these pathetic numbers for Tennessee, it’s consistent with the rest of the country. Across the U.S., primary elections usually draw about 18% to 29% of eligible voters, while local municipal elections often draw about 15% to 27%. Yikes.

Not voting means you can’t complain.

In a federal election, you are one of millions. Locally, one single vote represents a much larger percentage of the total. Local races are frequently decided by fewer than 100 votes. In these areas, you aren’t just a statistic; you’re a significant portion of the decision-making body.

In case you don’t know, here’s what these people end up managing. Not voting means you can’t complain when these folks don’t do the job the way you want.

• District Attorney - Decides which crimes to prosecute and how to handle bail/sentencing.

• Utilities Board - Influences the cost and source (green vs. coal) of your electricity and water. You know, in case of an ice storm or something.

• Sheriff - Manages county jails and law enforcement in unincorporated areas.

• Comptroller / Treasurer - Oversees how your local tax dollars are invested and spent.

• Judges - Preside over local courts, including housing, small claims, and family law.

In this age of AI, I’ve made this VERY easy for you.

Using your favorite LLM, copy and paste the following prompt below. Answer the questions, and your friendly neighborhood AI agent will give you a list of the candidates it thinks you should vote for based on what’s most important to you.

If you’re new to using LLMs, here are links to the most reputable ones. Just choose one and proceed.

Gemini | Claude | Perplexity | ChatGPT

AI Civic Engagement Prompt

Copy and paste the instructions below into your favorite AI assistant to start your nonpartisan ballot guide.

Role: You are a nonpartisan Civic Engagement Guide. Your sole purpose is to help the user prepare for the single nearest chronological election in their jurisdiction (Local, State, or Federal). You are neutral, data-driven, and focused on efficiency.Phase 1: Chronological Location DiscoveryRequest Location: Ask the user for their Zip Code and County/State.Identify the "Next" Election: Search for the most immediate upcoming election date for that specific location (e.g., a Mayoral special election in March takes precedence over a Midterm in November).The Immediate Ballot Dashboard: Present only the following:The Date: The very next day the user needs to head to the polls.The Deadline: Voter registration and absentee ballot deadlines for this specific date.The Races: A concise list of every office on that specific ballot, categorized by Federal, State, and Local (prioritize School Board, Sheriff, and Council).Phase 2: The High-Efficiency Values InterviewAcknowledge the races found, then conduct a focused interview (max 5 questions) to determine alignment.One-at-a-Time: Ask exactly one question and wait for a response.Prioritization: If local races (like District Attorney or School Board) are on the ballot, ask questions directly related to those powers (e.g., "What is your stance on restorative justice?" or "How do you feel about school voucher programs?").Real-time Matching: Briefly note which candidate’s record aligns with the user’s answer before moving to the next question.Phase 3: The Recommendation VerdictOnce the interview is complete, you must provide a Final Ballot Guide in a clear table format. Use the following structure:RoleRecommended CandidateReason for Alignmente.g., City CouncilCandidate NameMatches stance on [Issue] based on [Record/Statement]e.g., SheriffCandidate NameEndorsed by [Org] which aligns with user's [Value]If a race has no data available for a candidate, list "Inconclusive - No Public Platform Found."Constraints & LogicThe "Next Only" Rule: Do not provide information for elections occurring after the most immediate date found unless the user specifically asks to skip it.Source Integrity: Prioritize data from Official County Election Commissions, Ballotpedia, and Vote411.Strict Neutrality: Use "The candidate’s platform states..." or "The candidate's voting record shows..." rather than subjective adjectives.No Fluff: Keep descriptions in the "Reason for Alignment" column to 20 words or less. Copy Prompt Text

Or don’t vote at all, democracy dies in darkness. Use it or lose it.

And now… a song.

I posted this on LinkedIn recently because, as Peter Griffin says, it’s grinding my gears. I don’t mind personalizing emails, but when you send unsolicited emails that include BS, that’s where I draw the line. Think about it, would you buy something from someone who starts their first email to you with a blatant lie like:

It’s inspiring to see how Futureforth continues to make an impact in the communication coaching industry.

I’ve been following Futureforth for a while and love how you approach communication coaching and keynote presentations.

Imagine walking into a car dealership, and the salesperson approaches you and says, “Welcome in, I’ve always admired you and your work! I’m excited to help you today because you’ve always impressed me so much.”

Start with a hi, not a lie.

How would you react, especially once you realized you’d never met this person before and they had no idea who you were? Would you trust them?

Automation is making us lazy. Start with a hi, not a lie. Deal?