There's a particular kind of exhaustion that sets in when every tool, app, product, and platform is suddenly "powered by AI." When your calendar app wants to summarize your own calendar back to you. When a customer service bot answers your question with five paragraphs and zero information. When someone at a meeting describes a problem that could be solved with a sticky note, then announces they're going to "build an AI workflow" around it.

You are not wrong to be tired of this. That exhaustion is calibrated.

This guide exists because the conversation around AI has collapsed into two camps: the true believers who see a revolution in every chatbot, and the dismissers who refuse to engage entirely. Both are missing something. The believers are ignoring real costs and real limitations. The dismissers are leaving genuinely useful tools on the table out of principle.

This is the guide for everyone in the middle — the people who just want to get work done, hate wasting time, and are open to using whatever actually works. Even if, occasionally, that thing is Claude.

WHAT WE ACTUALLY BELIEVE

A simpler tool is always better when it works. If a calculator, a search engine, a template, or a spreadsheet formula solves the problem — use that. Faster, cheaper, and you don't have to describe your situation to a robot.
AI has a real environmental cost. Not catastrophic, not a reason to never use it — but real. A single AI query uses roughly the energy of a small light bulb for a minute and a sip's worth of cooling water. Scale that to millions of queries and the numbers matter.
Dependency is a design choice. Tools that make you dependent on them to do basic tasks are not empowering you. The goal of every tool in this guide — including Claude — is to leave you more capable, not less.
Context is everything. AI is genuinely extraordinary at a specific category of tasks: synthesizing complex information, drafting long-form content, explaining things you don't understand, and working through problems with no obvious right answer. That category is smaller than the hype suggests — and larger than the skeptics admit.
You should stay in control. Using AI well means knowing when to use it, what to give it, and when to stop. It means checking the output. It means owning the result. The moment AI starts making decisions you should be making, it's time to zoom out.

WHAT THIS GUIDE IS NOT

—
A pitch for AI adoption. We're not trying to convert you.
—
An AI horror story. We're not trying to scare you either.
—
A technical manual. No API keys, no jargon, no setup required.
—
Sponsored by Anthropic. This guide will tell you, plainly, when Claude is the wrong tool.
—
For developers. This is for people who work with documents, emails, decisions, and other humans.

By the end of this guide, you'll have a clear framework for when a low-tech tool beats AI every time, when AI genuinely earns its keep, and how to use Claude without becoming the person who asks it what to have for lunch.

The goal isn't to make you an AI user.
The goal is to make you a better judge of tools.

ONTOMATON · NEWBRAND.US

PART 1

The Math of Overkill

AI isn't software. It's infrastructure. Every time you ask Claude something, a request travels to a data center, spins up compute, draws power, and generates heat that has to be cooled with actual water. This is not a moral argument — it's just physics. And physics has opinions about using a sledgehammer on a thumbtack.

SIDE-BY-SIDE COST COMPARISON

CALCULATOR APP

What's 15% of $47?

0.001WATT-HOURS

~0ML WATER

<1SECOND

ASKING CLAUDE

What's 15% of $47?

~3WATT-HOURS

~100ML WATER

3–8SECONDS

GOOGLE SEARCH

What time does the post office close?

0.3WATT-HOURS

~3ML WATER

<2SECONDS

ASKING CLAUDE

What time does the post office close?

~3WATT-HOURS

~100ML WATER

5–12SECONDS

EXCEL / GOOGLE SHEETS

Sort this list of 200 names alphabetically

0.005WATT-HOURS

~0ML WATER

<1SECOND

ASKING CLAUDE

Sort this list of 200 names alphabetically

~8WATT-HOURS

~250ML WATER

10–30SECONDS

THE OVERKILL RATIO

3,000x

MORE ENERGY than a calculator

1,600x

MORE ENERGY than a Google search

100ml

WATER per average AI query

None of this means AI is bad. It means AI is expensive — and expensive tools have a place. You wouldn't use a crane to move a chair. The question isn't whether to use AI. It's whether the job is actually crane-sized.

THE RIGHT TOOL FOR THE RIGHT JOB — THE ESCALATION LADDER

Simple calculation, unit conversion, timer

Phone calculator / clock app

0.001 Wh

Look up a fact, address, hours, price

Google search

0.3 Wh

Sort, filter, or calculate with data

Spreadsheet (Excel, Sheets, Numbers)

0.005 Wh

Repeat a task you do every week

Template, macro, or newbrand.us tool

0.001 Wh

Draft a long document, analyze a complex situation, research something with no clear answer

Now you can use Claude

3–50+ Wh

Energy and water figures are estimates based on published research into LLM inference costs (University of Massachusetts Amherst, 2023; Goldman Sachs Energy Report, 2024) and Google's published per-search energy figure. Actual costs vary by model size, query complexity, data center efficiency, and cooling method. The point is the order of magnitude, not the decimal place.

PART 2

Your Phone Can Already Do That

Before we get to AI, let's establish something: the device in your pocket is extraordinary. It has a calculator, a search engine, a camera scanner, a unit converter, a timer, a weather station, a notes app, a spreadsheet, and about forty other tools that work instantly, offline, and without consuming a data center's worth of electricity. Most of what people use AI for, they didn't need AI for.

CALCULATIONS & CONVERSIONS

Never needs AI

What's 18% tip on $63.40?

Calculator app

63.40 x 1.18. Done in four taps.

How many miles is 14 kilometers?

Google / Siri / Spotlight search

Type "14 km in miles" into any search bar.

What's today's exchange rate?

Google search

Also: XE.com. Real-time, authoritative.

How many days until my deadline?

Calendar app / timeanddate.com

Your calendar already has this. Open it.

INFORMATION LOOKUPS

Google is faster

What are the store hours?

Google search

Hours appear in the knowledge panel instantly. AI has to look this up anyway — and may get it wrong.

What's the weather this weekend?

Weather app

Built into every phone. Uses live data. Claude cannot do this without a web tool enabled.

What does this word mean?

Dictionary app / Google

Type the word. Get the definition. AI adds a paragraph of context you didn't ask for.

Track my package

Carrier app / Gmail auto-track

Claude cannot track packages. Gmail can. So can every shipping app ever made.

SPREADSHEET TASKS

Excel was built for this

FORMULAS THAT REPLACE AI FOR 80% OF OFFICE MATH

Sum a column

=SUM(A1:A100)

Average without outliers

=AVERAGEIF(A1:A100,"<1000")

Count how many times X appears

=COUNTIF(B1:B50,"Yes")

Find a value in another table

=VLOOKUP(D2,Sheet2!A:B,2,0)

Combine first and last name

=A2&" "&B2

Flag overdue dates

=IF(C2<TODAY(),"OVERDUE","OK")

Don't know a formula? Type your question into the formula bar in Google Sheets — it has built-in AI assistance that runs locally and doesn't require a separate tool.

THE HONEST COMPARISON — GOOGLE VS. CLAUDE

USE GOOGLE WHEN YOU NEED

A specific fact with a known answer

Current information (prices, hours, news)

A direct link to a source

Images, maps, or local results

Something you can verify in one click

USE CLAUDE WHEN YOU NEED

To understand something, not just find it

To draft, rewrite, or structure text

To think through a decision or problem

To synthesize many sources into one answer

When the answer requires judgment, not lookup

FREE TOOLS AT NEWBRAND.US THAT REPLACE AI FOR SPECIFIC JOBS

Invoice Generator

Create and send professional invoices in under two minutes. No account required.

newbrand.us/invoice

QR Code Maker

Generate a QR code for any URL, text, or contact. Download instantly.

newbrand.us/qr

Email Signature Builder

Drag-and-drop your details. Copy the HTML. Paste into Gmail or Outlook.

newbrand.us/signature

Answer ten questions. Get a compliant privacy policy for your site or app.

newbrand.us/privacy

Business Name Checker

Search name availability across LLC registrations and domain names simultaneously.

newbrand.us/names

PDF Merge & Split

Combine or split PDFs in your browser. Nothing leaves your device.

newbrand.us/pdf

You've now got the full low-tech arsenal. So when does the ladder actually require climbing?

PART 3

The Escalation Ladder

Every task sits somewhere on a spectrum. At one end: things with a single correct answer, retrievable in seconds, requiring no interpretation. At the other end: things with no correct answer, requiring synthesis, judgment, and language. AI earns its cost at the second end of that spectrum. The ladder tells you where your task actually sits.

THE RIGHT TOOL, RIGHT JOB FRAMEWORK

STAY LOW-TECH WHEN

The answer exists somewhere and you just need to find it

The task is the same every time you do it

A formula or filter can do the work

Speed matters more than nuance

The source needs to be citable

You need real-time data (prices, weather, hours)

ESCALATE TO CLAUDE WHEN

There's no single right answer — only better or worse ones

The task would take you 30+ minutes to do manually

You need to produce language, not just retrieve it

You're synthesizing information from multiple sources

You want a first draft, not a final answer

The task requires understanding context, not just matching keywords

Real tasks, real verdicts

Every example below represents something a real person actually tried to use AI for. Some are obvious mismatches. Others are genuinely good uses. The verdict is based on the framework above.

LOW

"What's the capital of Australia?"

USE GOOGLE

Google search → 0.3 Wh

Single correct answer. Available instantly. AI adds latency and cost for zero additional value.

LOW

"Set a timer for 25 minutes"

USE SIRI / ALEXA

Voice assistant → ~0.001 Wh

This is what voice assistants were built for. Asking Claude to set a timer is like using a forklift to open a door.

LOW

"Sort this list of 50 client names alphabetically"

USE SPREADSHEET

Excel / Sheets: Data → Sort → ~0.005 Wh

Spreadsheets process structured data in milliseconds. AI will sort it correctly — after you paste it, wait for it, and copy it back out.

"Help me write a difficult email declining a longtime client"

USE CLAUDE

Claude → worth every watt

No formula does this. Tone, relationship history, desired outcome — this requires language, judgment, and craft.

"I have three job offers. Help me think through which one to take."

USE CLAUDE

Claude → genuinely useful

Complex tradeoffs, personal values, long-term implications. This is exactly the kind of open-ended synthesis AI handles well.

"Summarize this 40-page report into five key findings"

USE CLAUDE

Claude → hours saved, legitimately

Reading compression at scale is one of AI's clearest wins. Two minutes instead of ninety, with reasonable accuracy.

THE TIPPING POINT TEST — FIVE QUESTIONS

Is there a single, verifiable correct answer to this?

If yes → Google it

Could a formula, filter, or sort function handle this?

If yes → use your spreadsheet

Have you done this exact task before?

If yes → build a template

Does the answer require judgment, synthesis, or language?

If yes → Claude earns it

Would doing this manually take more than 30 minutes?

If yes → Claude earns it

Two or more "earns it" answers = justified. One or zero = reach for the low-tech tool first. You can always escalate if it's not enough.

You now know when to climb the ladder. Next: what to actually say when you get there.

PART 4

Claude for the Skeptic

You've made it to the AI section. The fact that you're still here means one of two things: you're genuinely open to being persuaded, or you want to know what you're arguing against. Either is fine. This section will not oversell. It will show you exactly how to use Claude — on your terms — and be equally honest about where it falls flat.

HOW TO PROMPT LIKE A SKEPTIC

Most people get mediocre results from AI because they ask vague questions and expect magic. Skeptics are actually better positioned here — you already don't trust it, which means you'll be specific, critical, and in control. That's exactly the right posture.

Be specific. Vague in, vague out.

Claude cannot read your mind. The more context you give — your role, your audience, your constraints, your goal — the better the output. Treat it like briefing a contractor, not wishing on a star.

WEAK

"Write me an email."

STRONG

"Write a two-paragraph email to a client who missed a payment deadline. Tone: firm but not hostile. We want to preserve the relationship. Include a specific deadline of Friday."

Tell it what you don't want.

Negative constraints are just as useful as positive ones. If you don't want bullet points, say so. If you don't want it to be overly formal, say so. Claude defaults to safe, generic, helpful — you can override every part of that.

WEAK

"Summarize this article."

STRONG

"Summarize this article in three sentences. No bullet points. Write it the way a smart colleague would explain it at lunch, not like a press release."

Ask for a draft, not a final answer.

The healthiest way to use Claude is as a first-pass generator, not a final authority. Ask for a draft. Read it critically. Edit it. The goal is to get 70% of the way there in two minutes, then spend your effort on the last 30%.

WEAK

"Write my resignation letter."

STRONG

"Give me a first draft of a resignation letter. I'll edit it heavily — just get me started. Two weeks notice, leaving on good terms, no need to explain why."

Push back when it's wrong.

Claude is not right by default. It will occasionally confabulate facts, misread your tone, or produce something generic. Tell it. "That's too formal." "You missed the point." "Start over with a different angle." It responds well to direct criticism.

WEAK

Accept the first output. Paste it into your email. Regret it later.

STRONG

"This is too stiff. Rewrite it — warmer, shorter, and cut the third paragraph entirely."

Always verify facts independently.

Claude's knowledge has a cutoff date. It can be confidently wrong about current events, specific statistics, and niche technical details. If the output contains a fact you're going to act on, check it. This is not a knock on AI — it's just how it works.

WEAK

"What's the current interest rate?" → use Claude's answer in a proposal.

STRONG

Use Claude for structure and language. Use Google or official sources for current facts, statistics, and legal specifics.

TEN TASKS CLAUDE IS GENUINELY GOOD AT

Not hype. Not demos. Real tasks that real people do every week, where AI saves meaningful time.

Drafting difficult emails and messages

"Write a message to my landlord about a repair that's been ignored for three weeks. Firm but professional. I want action, not conflict."

Tone calibration is hard. Claude is good at it. You edit the final version — it gets you unstuck.

Summarizing long documents

"Here's a 30-page lease agreement. Summarize the key obligations on my end, any unusual clauses, and anything I should push back on."

Reading compression at scale. Not a replacement for a lawyer — but a useful first pass.

Explaining complex topics simply

"Explain how a Roth IRA works to someone who has never invested before. No jargon. Use an analogy."

Patient, infinitely, without making you feel dumb. Better than most explainers online.

Preparing for a difficult conversation

"I need to give a performance review to an employee who's been underperforming but is well-liked. Help me think through what to say."

Helps you rehearse, anticipate objections, and find the right framing before you're in the room.

Editing and rewriting your own drafts

"Here's a proposal I wrote. Make it tighter — cut anything redundant, fix the structure, and make the opening stronger."

Editing your own writing is hard. An outside perspective, even an artificial one, helps.

Building a framework for a decision

"I'm trying to decide whether to hire a contractor or bring on a part-time employee. What questions should I be asking and what are the tradeoffs?"

Claude won't make the decision for you. But it will surface considerations you hadn't thought of.

Generating options when you're stuck

"I need ten ideas for a team offsite that aren't trust falls or trivia nights. Budget around $500. Team of eight."

Brainstorming with Claude is fast and low-stakes. You'll use two of the ten ideas. That's fine.

Translating technical content

"Here's a quote from a contractor for electrical work. Explain what each line item means in plain English."

Bridging the gap between specialist language and plain understanding. Consistently useful.

Writing job descriptions and interview questions

"Write a job description for a part-time bookkeeper at a small creative agency. Include five interview questions that reveal whether they're detail-oriented."

Structured writing tasks with clear parameters. Claude handles the blank page problem well.

Creating a first draft of anything long

"Draft an outline and introduction for a business proposal for a catering company pitching a corporate client. Formal tone, emphasis on reliability."

The hardest part of writing is starting. Claude removes that friction. The rest is yours to own.

WHERE CLAUDE ACTUALLY FALLS SHORT

UNRELIABLE

Current facts, prices, and live data

Knowledge cutoff means Claude can be wrong about anything that's changed recently. Never use it as your only source for statistics, current events, or anything time-sensitive.

UNRELIABLE

Specific legal, medical, and financial advice

Claude will always tell you to consult a professional — and it's right. It can help you understand concepts and prepare questions for your advisor. It cannot replace the advisor.

WEAK

Highly technical or niche domains

Claude knows a lot about a lot. But in specialized fields — advanced engineering, novel research, obscure legal jurisdictions — it will sometimes fill gaps with plausible-sounding nonsense. Experts should verify everything.

WEAK

Tasks requiring your personal voice

Claude can approximate your tone if you describe it well. But it will never sound exactly like you. For content that needs your genuine voice — personal essays, social media, your own brand — treat Claude as a starting point, not a ghost writer.

CANNOT DO

Anything requiring real-time internet access by default

Without web search enabled, Claude cannot check live prices, current news, or anything that happened recently. It's not connected to the internet the way Google is.

CANNOT DO

Make decisions for you

This sounds obvious, but it's worth saying: Claude can help you think. It cannot — and should not — replace your judgment. Every output is a draft. Every recommendation is an input. You are still the one deciding.

THE SKEPTIC'S RULES OF ENGAGEMENT

ALWAYS DO THIS

→

Read the output before you use it

→

Edit it to sound like you

→

Verify any facts that matter

→

Push back if it's wrong or generic

→

Own the final result

NEVER DO THIS

—

Paste AI output without reading it

—

Use it for tasks a simpler tool handles

—

Treat the first draft as the final answer

—

Let it make decisions you should make

—

Stop thinking critically because it sounds confident

You've got the mindset and the mechanics. Now let's put it to work — with the actual tools you use every day.

PART 5

The Office Anti-Suite

Microsoft Office didn't fail you. Google Workspace didn't fail you. You just never got a clear map of which tool handles which job — and where, at the exact right moment, it makes sense to hand off to AI. This section is that map. Start with the simplest tool that works. Escalate only when it doesn't.

COMMON OFFICE WORKFLOWS — WITH ESCALATION POINTS MARKED

Each workflow below shows the full task from start to finish. Steps with a coral marker are the escalation points — the moments where AI genuinely saves time. Everything else stays low-tech.

Sending a proposal to a new client

From blank page to sent document

WEEKLY TASK

Draft the proposal structure and opening paragraph

Claude — "Draft a proposal outline for [service] to [type of client]. Formal tone, one page."

~3 Wh

Fill in your specific pricing, timeline, and deliverables

Your own knowledge — no AI needed here

0 Wh

Format into a clean document

Google Docs or Word — use your saved template

~0.005 Wh

Review and tighten the language before sending

Claude — "Edit this for concision. Cut anything that sounds like filler."

~2 Wh

Export as PDF and send via email

Your email client — Gmail, Outlook, Apple Mail

~0.01 Wh

WITHOUT AI

MINUTES AVERAGE

Staring at a blank page, writing and rewriting the opening, second-guessing the tone.

WITH THIS WORKFLOW

MINUTES AVERAGE

AI handles the blank page and the final polish. You handle the substance. That split is the whole point.

Monthly expense reporting

From receipts to submitted report

MONTHLY TASK

Collect and photograph receipts throughout the month

Phone camera — file into a folder as you go

0 Wh

Enter amounts into your expense spreadsheet

Google Sheets — use =SUM() and category filters

~0.005 Wh

Write the narrative summary for your manager or accountant

Claude — "Here are my expense totals by category. Write a two-paragraph summary for my manager explaining the main categories and any unusual items."

~2 Wh

Generate and send the invoice if billing to client

newbrand.us/invoice — free, no account required

~0.001 Wh

Hiring a contractor or part-time employee

From need identified to first interview

OCCASIONAL TASK

Write the job description

Claude — "Write a job description for a part-time [role] at a small [type] business. Include responsibilities, requirements, and compensation range of [X]."

~3 Wh

Post to job boards and review applications

Indeed, LinkedIn, Craigslist — your call based on role

~0.3 Wh

Prepare interview questions

Claude — "Give me eight interview questions for this role that reveal attention to detail, reliability, and how they handle ambiguity."

~2 Wh

Conduct interviews and check references

In person or video call — this part stays human

0 Wh

Draft the offer letter or contractor agreement

Claude — "Draft a simple offer letter for a part-time contractor. Rate: $X/hr, start date: [X], scope: [X]. Have a lawyer review before sending."

~3 Wh

Handling a difficult customer complaint

From angry email to resolved situation

UNPREDICTABLE TASK

Read the complaint carefully. Understand what they actually want.

Your judgment — no tool replaces this step

0 Wh

Draft a response that acknowledges, doesn't over-apologize, and proposes a resolution

Claude — "A customer is angry about [X]. We can offer [Y]. Write a response that's empathetic but not groveling, and moves toward resolution."

~3 Wh

Edit the draft so it sounds like you, not a template

Your voice — this edit takes two minutes and matters

0 Wh

Send and follow up if needed

Your email client

~0.01 Wh

WHEN YOUR LOW-TECH TOOL HITS ITS LIMIT — ESCALATION MAP

WHAT YOU STARTED WITH

WHEN TO ESCALATE TO CLAUDE

Word / Google Docs draft

→

When you've rewritten the opening three times and still hate it

Spreadsheet with data

→

When you need to explain what the numbers mean in plain English

Email thread you need to reply to

→

When the situation is delicate and tone really matters

List of ideas you've brainstormed

→

When you're stuck and need ten more options fast

Template you've used before

→

When this particular situation is different enough that the template doesn't fit

Google search results

→

When you have the facts but need help understanding what they mean for your situation

The workflows are in place. Now let's talk about what you're actually sharing when you use any of this — and how to stay in control of it.

PART 6

AI Hygiene for Normal People

If you've made it this far, you're using Claude more thoughtfully than 90% of people who use it. But there's one more thing worth understanding: what actually happens to what you type. Not because it's a scandal — but because knowing it lets you make smarter choices about what to share and what to keep offline.

WHAT YOU'RE SHARING WHEN YOU USE CLAUDE

Everything you type into Claude is processed by Anthropic's servers. That's how it works — the model runs in the cloud, not on your device. Here's how to think about what's appropriate to share.

GENERALLY FINE TO SHARE

General business descriptions
Anonymized situations and scenarios
Document drafts without personal info
Industry questions and research
Creative writing and brainstorming
Public information you're trying to understand

SHARE WITH CARE

Client names and details — anonymize where possible
Internal financial figures
Employee performance details
Sensitive business strategy
Legal matters under active dispute
Anything under NDA

KEEP THIS OFFLINE

Passwords or login credentials
Social security numbers
Full credit card or bank details
Medical records with identifying info
Classified or regulated information
Other people's private communications

FIVE MYTHS ABOUT AI PRIVACY — CORRECTED

MYTH

"Claude remembers everything I've ever told it."

By default, Claude does not retain memory between separate conversations. Each new conversation starts fresh. There is a memory feature you can enable — but it's opt-in, and you can view and delete what it stores at any time in your settings.

MYTH

"Anthropic is reading my conversations to sell me things."

Anthropic does not sell user data or run an advertising business. Conversations may be reviewed by staff for safety and model improvement purposes — which you can opt out of in your account settings under data privacy.

MYTH

"If I use Claude for work stuff, my company's data is exposed."

Using the consumer product (claude.ai) for sensitive business data is worth thinking about carefully. Claude for Teams and Enterprise have stronger data handling commitments — conversations are not used for training and data stays within your organization's agreement.

MYTH

"AI is always listening, like a smart speaker."

Claude is not ambient. It only processes what you actively type and submit. Nothing is captured passively. It has no access to your microphone, camera, files, or anything outside the conversation window unless you explicitly upload or paste it.

MYTH

"Deleting a conversation means it's gone forever."

Deleting a conversation removes it from your visible history. Anthropic may retain logs for a period of time for safety and legal compliance purposes, as outlined in their privacy policy. For truly sensitive material, the safest approach is not to share it at all.

SETTINGS WORTH KNOWING ABOUT

Conversation history

claude.ai → Settings → Data & Privacy

You can delete individual conversations or your entire history. Useful if you've shared something sensitive and want to clear the record.

Recommended: review monthly

Training data opt-out

claude.ai → Settings → Data & Privacy

You can request that your conversations not be used to train future models. This is available on paid plans and through the privacy request form.

Recommended: enable if using for business

Memory settings

claude.ai → Settings → Memory

If memory is enabled, Claude builds a profile of what you've told it across conversations. You can view, edit, or delete any stored memory at any time.

Recommended: review what's stored quarterly

Account deletion

claude.ai → Settings → Account

Full account deletion removes your profile and associated conversation history. Anthropic will confirm the deletion timeline in their response.

Nuclear option — available if you need it

THE AI HYGIENE CHECKLIST

BEFORE YOU TYPE

Would I be uncomfortable if a colleague read this conversation?

Does this contain real names, account numbers, or sensitive identifiers I can anonymize?

Is this covered by an NDA or confidentiality agreement?

BEFORE YOU USE THE OUTPUT

Have I read the full response before pasting it anywhere?

Have I verified any facts, statistics, or dates it mentioned?

Does this actually sound like me, or like a chatbot?

Am I comfortable owning this output as my own work?

PERIODICALLY

Clear old conversations that contain sensitive details

Review your memory settings if enabled

Ask yourself: am I using AI because it helps, or out of habit?

None of this is meant to frighten you. Anthropic is a safety-focused company with a published privacy policy and genuine commitments to responsible AI. The hygiene practices above are just good data practices for any cloud service — the same common sense you'd apply to Google Drive, Dropbox, or email. The goal is informed use, not paranoia.

That's the full guide. All that's left is the one page you'll actually keep.

APPENDIX

The Quick Reference Card

Everything in this guide, on one page. Keep it. Share it. Argue with it.

ALWAYS USE A SIMPLE TOOL FOR

—Calculations and unit conversions
—Current facts, prices, and hours
—Weather, timers, reminders
—Sorting and filtering structured data
—Generating invoices and QR codes
—Anything with a single correct answer
—Tasks you do the same way every time

ESCALATE TO CLAUDE WHEN

→There is no single correct answer
→You need to produce or improve language
→The task would take 30+ minutes manually
→You need to synthesize multiple inputs
→Tone and judgment matter
→You're stuck and need a first draft
→You need 10 options fast

THE FIVE-QUESTION TIPPING POINT TEST

STAY LOW-TECH

Is there a verifiable correct answer? → Google it

Can a formula or filter handle it? → Use your spreadsheet

Have you done this exact task before? → Use a template

CLAUDE EARNS IT

Does it require judgment, synthesis, or language?

Would doing it manually take 30+ minutes?

—

Two "earns it" answers = justified escalation

PROMPTING IN SIX MOVES

Be specificInclude your role, audience, tone, and constraints

Say what you don't wantNo bullet points, no formal tone, cut the fluff

Ask for a draftNot a final answer — a starting point you'll edit

Push back"Too stiff." "Missed the point." "Try again."

Verify factsAny statistic or date you'll act on — check it

Own the outputEdit it to sound like you before using it

TEN TASKS CLAUDE IS GENUINELY GOOD AT

01Drafting difficult emails
02Summarizing long documents
03Explaining complex topics simply
04Preparing for hard conversations
05Editing and tightening your drafts
06Building decision frameworks
07Generating options when stuck
08Translating technical language
09Writing job descriptions and questions
10Creating first drafts of anything long

WHERE CLAUDE FALLS SHORT

—Current facts, prices, live data
—Specific legal or medical advice
—Highly specialized technical domains
—Your genuine personal voice
—Real-time internet access by default
—Making decisions for you
NEVER SHARE WITH ANY AI
—Passwords or credentials
—SSNs, banking details, medical records
—Anything under NDA

Claude for Haters — The Anti-AI User's Guide to Claude · First Edition

newbrand.us

You finished the book.

Now you know when to use a calculator, when to use a spreadsheet, and when Claude actually earns its keep. Share it with someone who needs the framework.

Download the PDF Visit newbrand.us

← Back to newbrand.us