Jumping off a Cliff
By Zeel Jain
FIELD GUIDE - JUST START
Stop Learning AI. Start Using It.
Four real problems you are already facing this week - and the prompts that turn each one into a ten-minute move with Claude or ChatGPT.
4
Moves you can run today
10 min
To the first useful output
0
Tutorials required
Preparation is the most expensive form of procrastination.
The gap between people getting ahead with AI and people watching them is not intelligence or technical skill. It is the willingness to open the tool and give it a real problem from this week.
There is a loop almost everyone gets stuck in: "I need to learn more before I start." Another course. Another tutorial. Another saved thread of prompts you will get to on Sunday. None of it is preparation. It is fear wearing the costume of progress.
Courses teach you a tool in the abstract. Abstract knowledge does not compound. What compounds is the moment you give the tool an actual decision you owe by Friday - a quote, an email, a model, or a call - and see what comes back. That output is usually seventy percent there. The remaining thirty percent is yours to add. Doing is learning.
The shift: stop reading about AI on the train and start running it on the problem you have been avoiding since Monday. The first output is rarely the finished thing. It is the head start that finally lets you finish.
MOVE 01 · PRICING
The quote you've been avoiding.
Brief in, defensible price out.
"What should I charge?" consumes more founder and freelancer hours than the work itself. The number sits in a draft for three days while you ask social media, scroll through a pricing course, and stare at your last invoice for clues. The client is waiting for one email, and you are losing the work to the person who replied that evening.
The shortcut: paste the full brief, including the scope, your experience, the client's stage, and the timeline. Ask for a defensible range, the questions to send before quoting, and the line that justifies the top of the range. You get a quote-ready answer in the time it takes to make coffee.
Example: BRAND IDENTITY FOR A FINTECH, ~6 WEEKS
You ask Claude: "I am a brand designer with five years of experience. I have been asked to redesign a Series A fintech company's identity - logo, type system, and guidelines - over approximately six weeks. They want a quote. Help me price it without lowballing or scaring them off."
STRENGTHS
- It ends analysis paralysis by giving you a range rather than a guess.
- It surfaces the qualifying questions you forgot to ask before quoting.
- The explanation for the top tier becomes language you can use in your pitch.
LIMITS
- It does not know your local market or your reputation premium.
- It cannot read the client's body language. That remains your job.
- The numbers need a final review before they go into the email.
Use this when: a quote is due and you have been staring at a blank email for more than a day.
MOVE 02 · OUTREACH
The email you've rewritten six times.
Context in, sendable email out.
Cold emails, follow-ups, and hard conversations die in the draft folder. It is not because you do not know what to say; the blank page is simply the worst place to start. You rewrite the first sentence six times, close the tab, and promise to finish tomorrow. The thread cools, and the opportunity drifts.
The shortcut: give AI everything you would tell a colleague before they wrote the email for you: the recipient, the history, the request, and what you want the person to remember. You get a clean draft, two subject options, and a recommended send window. Edit for voice, send it, and continue with your day.
Example: A COLD WARM EMAIL TO A HEAD OF DESIGN
You ask Claude: "I need to email the Head of Design at a Series A company. We met briefly at a meetup six weeks ago. I want to pitch a one-week brand-audit consultation. Write the first email."
STRENGTHS
- It removes the blank-page problem by giving you something to react to.
- It structures the request clearly: recall, value, and opt-out.
- It can suggest subject lines you may not write under pressure.
LIMITS
- The voice needs your edit pass because the first draft can sound generic.
- It does not know what has already been said. Give it the previous thread.
- Hard conversations still need your tone on top of the structure.
Use this when: three days have passed and the message you owe is still in drafts.
MOVE 03 · NUMBERS
The model you keep starting over.
Inputs in, working projection out.
Financial models are ninety percent structure and ten percent judgment. The ten percent is yours. The ninety percent is the part you keep restarting at midnight because an empty spreadsheet does not tell you which cells matter. Most people never get past the column headers.
The shortcut: give AI your real numbers, including current revenue, margin, and the scenarios you are weighing, and ask for a twelve-month projection across the options. You get a side-by-side comparison, break-even points, and a short list of assumptions to pressure-test. The structure is solved. Your judgment goes into the assumptions, not the layout.
Example: 4- PERSON STUDIO CONSIDERING TWO SENIOR HIRES
You ask Claude: "I run a four-person design studio with Rs 14 lakh in monthly revenue and approximately 35% margins. I am considering hiring two senior designers at Rs 1.1 lakh each. Build a twelve-month projection for three scenarios: hire now, hire in March, or hold."
STRENGTHS
- It ends the empty-spreadsheet stare by producing a structure in minutes.
- It surfaces scenarios you may have forgotten to consider.
- The pressure-test list resembles the one a good chief financial officer would prepare.
LIMITS
- It can hallucinate tax and regulatory details. Verify those separately.
- Assumptions are guesses dressed as confidence. You must own them.
- Garbage in, garbage out. Your inputs need to be real.
Use this when: a decision depends on numbers - a hire, a price change, or a runway call - and you cannot picture the next twelve months.
MOVE 04 · JUDGMENT
The decision you've been postponing.
Context in, framed decision out.
Most statements that begin with "I need more information" actually mean "I am scared to decide." The information you have already gathered is usually enough. What is missing is the frame: a way to lay out the choice so the right question becomes obvious. That is the part you keep delaying because it requires sitting with discomfort.
The shortcut: give AI the full context, the options, and your instinctive preference. Ask it to map the trade-off, surface the questions you have not asked, and flag what you are avoiding. It will not make the decision. It will get you to the decision faster, which was the bottleneck all along.
Example: JOB OFFER, FOUNDER WEIGHING A HEAD-OF-DESIGN ROLE
You ask Claude: "I have been offered a Head of Design role at a Series B SaaS company: Rs 38 lakh in cash plus 0.4% equity. I currently run my own studio, generating Rs 18 lakh in monthly profit. I am 33, have two children, and have six years of runway. Help me think this through."
STRENGTHS
- It maps blind spots and surfaces the questions you have been avoiding.
- It gently counters your preferred narrative.
- It names the question your instinct already had but could not express.
LIMITS
- It does not know your risk tolerance. Only you do.
- It cannot replace a conversation with your partner or co-founder.
- The final decision is still yours. The frame is the gift, not the answer.
Use this when: you have been "thinking about it" for more than a week and the thinking is no longer producing new information.
The four moves at a glance
How to start.
Pick the move already nagging you. Do not choose the one with the highest leverage on paper. Choose the one you have already half-drafted in your head this week. Run it on the actual context. Do not sanitize the inputs or pretend the problem is hypothetical. The first output is a head start, not the finished thing.
The first output will be about seventy percent complete. The remaining thirty percent is what makes it sound like you rather than the model. That editing layer is where your judgment lives, and it is the only part of the work you cannot outsource. Notice what you change. Those changes are the signal.
Do not bookmark this and promise to return tomorrow. The people who pulled ahead with AI did not have a better tool or a smarter prompt. They opened it, worked through a real problem, and now cannot imagine working without it.
Close this tab. Open Claude. Give it something real.