AI won’t leave you behind. Your questions will.

A free 8-page guide to the one skill Elon Musk named when he was asked how a 20-year-old avoids being left behind. You can start practising it tonight.

Send me the free guide (PDF)

Here is the uncomfortable part.

The tools are already good enough.

The person next to you has the same chatbot you do.

Same model. Same price. Same blinking cursor.

So why does one of you get back a finished table with a source for every number…

…and the other gets ten generic bullet points that thousands of other people got too?

It isn’t the tool.

It’s the ask.

And there is one line you can add to the end of almost any ask that catches more AI mistakes than anything else. It takes a single sentence. I’ll show you exactly what it is in a moment.

First, why this matters more every month.

Even Elon Musk says this is a question problem

In a September 2026 interview with China Media Group, Musk was asked how a 20-year-old can avoid being left behind during the AI transition.

He didn’t say “learn to code.” He didn’t name a tool. At about 23:12 in the video, he said:

“In order to know what to ask the robots for, you need to be able to formulate the question.”

Source: Full: Exclusive CMG interview with Tesla CEO Elon Musk, CGTN, YouTube, https://www.youtube.com/watch?v=34-5wOGmCKc

Read that again. The bottleneck isn’t the robot.

It’s you, the moment you open your mouth.

What happens if you never learn this

You ask for help.

You get mush.

You decide AI is overhyped, and you go back to doing it all by hand.

Meanwhile someone who writes clearer jobs hands off the same work, checks it in five minutes, and moves on to the next thing.

Then does it again tomorrow. With two bots. Then five.

My bet, and it is an opinion, not a statistic: most people will wait for someone to tell them what to learn next. A smaller group will practise giving AI clear jobs, and they will end up with real leverage.

This guide is for the second group.

What’s inside the free guide

The five-part job formula that turns “help me with this” into work you can actually use. You can write it in a few sentences.
The one-sentence line that makes the AI confess what it doesn’t know, instead of quietly filling the gap with a guess.
Why the longer ask is the faster one, and the three-minute habit that saves an hour of back and forth.
Three before-and-after asks, from a student, a founder and an operator, so you can see the difference line by line.
The rule for running several bots at once without them waiting on each other: split by output, not by step.
The Day 3 exercise, where most of the skill is actually built. (It is not the day you would guess.)
11 free field guides from SpaceXAI, matched to your job, so you know exactly where to start.

Get the guide free (PDF)

The chain that decides who gets left behind

Broad knowledge
The right question
A team of bots
Done

Break the second link and nothing after it works. That is the link this guide fixes.

Five parts that turn a vague ask into finished work

Every good job you give an AI has five parts. Skip one and you will usually get back something generic, wrong, or unusable.

1. Goal. What do you want, and why? The why changes everything.
2. Context. What does the AI need to know that it can’t guess?
3. Constraints. The limits: budget, length, deadline, tools allowed, things it must not do.
4. Success criteria. How you will judge the result. Concrete enough to check in five minutes.
5. Report back. What you want to see at the end, and in what shape. Including the line I promised you: “Tell me what you could not verify.” It is the cheapest way there is to catch mistakes before they reach you.

Easy to remember: Goal, Context, Constraints, Criteria, Report.

Watch what it does to one real ask

A student picking a thesis topic.

The ask most people write:

Give me thesis ideas about AI.

Result: ten generic topics that thousands of other students are also getting.

The ask that gets used:

Goal: I need a master’s thesis topic in economics that I can finish in eight months and that would interest a fintech employer.

Context: I’m at a university in Bratislava. I know Python and basic econometrics. My supervisor works on labour markets. I have access to Eurostat and Slovak statistics office data.

Constraints: Only topics that can be done with public data. No topics that need a lab or surveys of more than 100 people.

Success criteria: Five topics. For each: the research question in one sentence, the dataset, the method, and one recent paper (with link) that is close but not the same.

Report back: Rank them by how feasible they are in eight months, and flag any where you could not find the dataset.

Same student. Same tool.

The first ask gets a list. The second gets something you can walk into your supervisor’s office with.

The guide has two more of these, for a founder researching competitors and an operator clearing a support backlog.

Show me all three (free PDF)

Already know your field? Skip straight to your guide

Free guides from SpaceXAI on using Grok Bot for specific kinds of work. Start with the first one, then pick the one closest to what you do.

Getting started. For anyone who has never handed real work to a bot.
Engineering. For developers who want several engineer bots working on code in parallel.
Mobile app development. For solo builders or small teams making an app or game with a set of bots.
Design. For designers who want to explore more ideas than they have time to sketch.
Product management. For PMs who want research, specs and follow-ups handled by a team that reports to them.
Marketing. For marketers researching competitors and drafting positioning.
Founder-led sales. For founders doing their own prospecting, research and outreach drafts.
Recruiting. For hiring managers and recruiters doing interview prep, scorecards and sourcing.
Legal ops. For in-house legal and ops people triaging contracts and preparing for meetings.
Support. For support leads who want to work through the whole queue with bots.
Templates. For anyone who built a bot setup that works and wants to share it as a reusable recipe.

Seven days. Real work. No toy examples.

Thirty to sixty minutes a day. By the end of the week you have a backlog, a few reusable templates, and a clear map of where your asks are weak.

Day 1. Write down your backlog.
Day 2. One job, five parts.
Day 3. Rewrite and rerun. (This is the one that matters most.)
Day 4. Read your field’s guide.
Day 5. Two bots at once.
Day 6. Learn something outside your field.
Day 7. Save your best job as a template.

The full instructions for every day are in the guide.

The work is coming either way. Be the one who gives the orders.

AI will do more of the work every year. That part is not up to you.

What is up to you is whether you are the person who knows what to ask for, can tell a good result from a bad one, and can keep a team of bots busy on things that matter.

That skill has two halves. Knowing enough about the world to ask good questions takes years and never really ends.

Turning a question into a clear job? You can start tonight.

Send me the free guide (PDF)

Free. 8 pages. Instant download. No email required.

[CTA: join / get the full product]

P.S. If you only take one thing from this page, take the line from part 5. Before your next ask, add: “Tell me what you could not verify.” Then grab the guide for the other four parts and the seven-day plan.