What the people who get the most out of AI do differently
ChatGPT and Claude are brilliant, but they only remember scraps of who you are, and rarely the things that matter. Second Self is a guided interview that turns your story and context into files your AI keeps, so it stops handing you a stranger's advice and starts giving you yours.
One-time payment · No new subscription · 30-day no-quibble guarantee
or see exactly what it changes ↓Sound familiar?
None of them is a disaster. That's what makes them so frustrating.
Generic advice
A tidy list written for anyone alive. Useful to nobody, least of all you.
Groundhog day
Job, kids, budget, typed in again. Like reintroducing yourself to a colleague every morning.
Wrong fit
A 5am routine, for someone on school runs and shifts. Quietly judged by a chatbot.
Wiped clean
One brilliant answer. Next week it's forgotten you, and you're back to square one.
So you ask smaller questions, and wonder why everyone else gets so much more from the same app.
It's not the AI. It's working blind.
The practice
The people who get the most from AI do one thing differently. They write a short brief about themselves, once, and their AI reads it before every conversation. Not a prompt, not a trick, not another subscription.
The rule underneath is simple: AI can only be as personal as what it knows about you. Give it nothing and you get the internet's average. Give it your story, and the same tool starts answering like someone in your corner.
“By default, the AI gives you answers from the center of the cloud, the most likely answers to the question for the average person… Since the AI only ‘knows’ the context it is given, working with it through a series of questions gives the AI context it can use to help you.”
Ethan Mollick · Professor at Wharton, author of Co-Intelligence
One Useful Thing, November 2023 · oneusefulthing.org
That brief is called context. Building one used to mean knowing what to include and how to structure it. Second Self turns it into a 30-minute guided interview. No jargon, no blank page, no tech skills.
Your clone is simply your own ChatGPT or Claude once it's read these files. Same app, same login, now briefed on you. Nothing to install, nothing hosted, and nobody else can talk to it.
When you finish, you get your context pack:
About Me
Your story, goals, constraints and how you think. The file that does most of the work.
About My Work
Optional: what you do, how the money works, and your edge, for a clone that talks business.
Who Your Clone Is
A job description for your AI: what to challenge you on, what never to do, how to sound.
Setup Guide
Plain-English installation for Claude and ChatGPT. Copy, paste, done in five minutes.
Each one downloads twice over: as Markdown, which is the format AI reads best, and as a PDF you can read yourself.
Set it up once, and this is what changes:
The difference, in one question
Five everyday questions, asked by four example people we built complete packs for: a gardener, a programme manager, a designer and an engineer. Four very different lives, and the same complaint every time: the answer it gave me could have been written for anybody. Pick one and watch what changes.
Nadia, 43
Programme manager · corporate
Twenty-one years of delivery, still called “a safe pair of hands”. Wants the job where she decides what, not just how fast.
This is a common crossroads, and there are a few things worth weighing up:
One useful approach is a conditional yes — accepting while negotiating for the support you need. It may be worth having an open conversation with your director about expectations and capacity.
Sensible, unobjectionable, and word-for-word what it would say to the person sitting next to her.
Yes — but not as it's currently worded, and not for free.
This is the third rescue job in eighteen months. You're superb at them, which is exactly why you keep being handed them and exactly why they still call you a safe pair of hands. Another quiet rescue does not get you to Director. A visible one might.
So say yes with two conditions, in writing:
And a warning from your own file: you have four non-meeting hours a week. If this arrives without something coming off, it arrives in the evenings — and you told me bedtime three nights a week isn't negotiable.
It knows she's going for Director, that she has four usable hours, that she absorbs other people's fires, and where her line is.
Sofia, 34
Freelance designer · young family
Fifteen working hours a week between nursery runs. Wants to choose the work, not take what comes.
Great — a free block is a real opportunity! A few suggestions:
Try picking whichever will make tomorrow easiest for you.
Six options and no answer. The deciding was the part she wanted help with.
Pitch. Two hours, three emails.
Not the website. It's been three years out of date and it will still be out of date on Friday — and you told me redesigning it is what you do instead of asking, because designing is comfortable and asking isn't.
You're two retainers short of Christmas with four months left, so roughly one conversation a fortnight has to start. Three obvious ones: the skincare client who's referred you twice, the deli whose packaging you did last spring, and the person from the Bristol event you keep meaning to follow up.
Want me to draft all three? Twenty minutes, and you'll still have ninety left.
One instruction instead of six options — because it knows her goal, her deadline, and the exact thing she does to avoid both.
Dana, 38
Landscape gardener · self-employed
Busy, skilled, and quietly worried his knees give out before the business can run without them.
Raising prices can feel daunting but is often necessary. Consider:
Many businesses find a 5–10% increase is well tolerated. Communicate the change clearly and confidently, and be ready to explain the value behind it.
Textbook — and completely unaware that Dana is drifting toward a tax threshold that makes this the wrong question this month.
Yes. But it isn't your problem, and I'd fix the other thing first.
Your day rate's £320, projects run £12k–£25k at about 35%, and you win most of what you quote. That isn't a pricing problem, it's a throughput problem — quoting eats four evenings a week, so you lose weeks of pipeline to your own bottleneck. Getting quotes out inside 48 hours is worth more than 10% on the rate.
Now the thing you didn't ask about. You're creeping toward the VAT threshold. Cross it and your prices effectively rise 20% overnight for domestic customers whether you've planned for it or not — and most of your work is domestic.
That's the number to model this month. Want me to run it three ways: stay under, cross it deliberately, or reprice now?
It answered the question, then raised the better one — because it knows his margins, his bottleneck and the threshold he's drifting toward.
Ray, 57
Engineer · second act
Twenty-nine years at one firm, redundancy in March. Wants to choose the next ten years this time.
Here's a draft you could adapt:
Feel free to tailor it to the specific role or audience you have in mind!
Could belong to any engineer alive. Twenty-nine years, and not one of them is in there.
Not a summary. A sentence you'd actually say out loud.
Read this aloud and see whether it sounds like you:
Note what I left out: passionate, proven track record, seeking opportunities. You said you'd never put your name on something you'd be embarrassed to explain at the table — and that first paragraph is exactly that.
One more thing. You wrote down a worry that everything you know is specific to one company. Four of the people you trained now run their own teams elsewhere. That transferred.
It wrote in his words, cut the CV-speak he'd told it he hated, and quietly answered the fear in his file.
Dana, 38
Landscape gardener · self-employed
Busy, skilled, and quietly worried his knees give out before the business can run without them.
Protecting your time off matters for avoiding burnout. Some strategies that work well:
Consistency is key — start small and build the habit over time.
Advice for a person with no name. Nothing here knows why his Sundays actually disappear.
You won't fix Sunday on Sunday. You'll fix it on Wednesday.
Here's the actual mechanism. You say yes to a £400 patio repair for someone nice. That costs you the evening you'd set aside for quoting. The quotes slide to the weekend. Sunday goes. It is never a decision about Sunday — it's four small yeses on a Tuesday.
Two changes, and only two:
And something you told me rather than something I'm inventing: Maisie is nine. You get about four more years of her wanting to spend Sunday with you. That's the number that matters here — not the £400.
It found the cause rather than the symptom, gave him two rules instead of five, and used the one fact that will actually make him do it.
That is one question, once. Here is someone who could afford any adviser alive, and still thinks things through at length with AI.
“When I’m working on malaria, nutrition, my poor humans that I work with always get these long conversations from me, where I paste in — me, Claude, me, ChatGPT. Sometimes I do it if there’s three of us: Claude, ChatGPT and me, debating these things.”
Bill Gates
GeekWire, August 2026 · geekwire.com
How it works
01
25 min
Not a form. An interview. Seven short sections, and your draft saves as you go.
02
instant
Your complete context pack, generated the moment you finish. Yours to keep forever.
03
5 min
Paste the files into a Claude or ChatGPT project. Don’t worry, the Setup Guide walks you through it click by click. No coding, no new subscription.
Ray Dalio built an AI clone of himself from decades of his written principles and Bridgewater data. Read the second half of what he says a clone is for — it is the half almost everyone misses.
“I believe that I have a good idea of what [AI clones] have the potential to do for both a) those who use the digital clones to access the thinking of the individuals cloned and b) the individual being cloned who wants to use the digital clone as a thought partner.”
Ray Dalio · founder of Bridgewater Associates
Fortune, October 2025 · fortune.com · the square brackets are Fortune’s
The worry nobody says out loud
It's the fear that stops most people, so let's say it plainly: you can't do this badly. You're not being asked to be interesting, or to have your life worked out. Just what you do, who depends on you and how many hours you really have. You already know all of it. Nobody's ever asked you in the right order.
Hints, worked examples and four sample packs. You're never staring at an empty box.
Half sentences are fine. Your AI reads for meaning, not prose.
Your draft saves as you go. Come back tomorrow, or start again whenever life changes.
Answer the way you'd answer a friend who actually asked. The rest takes care of itself.
The offer
Seven sections, about twenty-five minutes, with questions sharp enough to do the thinking for you. Jump around freely and answer them in any order. Your draft waits on your device until you're done.
All four files, written from your answers, ready to paste into Claude or ChatGPT. You won't be left to work out where they go. The Setup Guide takes you through it click by click, in plain English. Keep them forever, and redo them whenever life changes.
There is no account to create and nothing is ever uploaded. The interview runs on your own device, so everything you type stays on it. We couldn’t read your answers even if we wanted to.
Twelve months from now, your clone writes you a letter about the year you have just had. It knows your goals, your constraints and the hours you actually have, so it doesn't write the flattering version. One rule makes it work: the letter must include a goal you missed, a cost you paid, and a worry that never quite resolved. Written to be believed, not to cheer you up.
Designed PDF guide · delivered instantly
A seven-day course in one designed PDF. One short exercise a day, ten minutes each, and every one puts your clone to work on something real rather than something clever. Argue it out of a decision you have already made. Fix the line it got wrong about you. Finally open the thing you keep moving down the list. By day seven you have a habit instead of a download.
Seven-day PDF course · delivered instantly
One-time payment · Instant access · 30-day no-quibble guarantee
A word from Mark
Hi, I'm Mark.
Eighteen years in a corporate finance role, CIMA-qualified, and I paid for AI for over a year before I admitted I was getting almost nothing out of it. (A bit of a slow learner.)
Not nothing as in broken. I'd ask it something that actually mattered and get back an answer written for a person who doesn't exist. No job, no family, and forty free hours a week to reorganise their life in.
So I'd explain myself. What I actually do all day. Who depends on me, and when. That I have two usable evenings a week rather than the five it keeps planning around. And for that conversation it was genuinely good, right up until I opened a new chat the next morning and had to do the whole thing again.
After a year of that, and a fair bit of time and money spent trying to make it work, the obvious finally landed. The AI wasn't the problem. I was handing it nothing to work with.
So I wrote it down properly, then spent weeks getting it wrong. Too much detail, too little, the wrong kind. What eventually worked became a clone that acted as my second self.
Here is what changed, and it changed within a day. It stopped suggesting things I'd never do. It started drafting in my words instead of a press release's. It did the arithmetic on a target I was rather attached to, told me it needed a hundred and one customers, and asked whether I had actually checked. Same subscription, same model, same £20 a month. The only thing that changed was what it knew about me.
When friends saw it working they didn't ask how it worked. They asked for one. Second Self is that method as a guided interview. Instead of the months and the money it cost me to work out, it takes thirty minutes, and it puts you ahead of almost everyone else paying for the same AI subscription, because almost nobody does this.
I won't promise it will change your life. I don't know you, which is precisely the problem this fixes. So I'll promise this instead. Half an hour from now, your AI could finally know whose life it's talking about.
— Mark
Founder, Second Self · a product brought to you by Simple Soul
Nothing you type ever reaches us.
There's no Second Self server, no account, and no database of your answers. The interview runs entirely in your browser, and your files live in your own AI subscription. We couldn't read your answers if we wanted to, and we built it that way on purpose.
The price
“I use ChatGPT almost every day — it’s developed a rich contextual memory of how I think, which makes it an incredibly effective partner for writing, decision-making, and pressure-testing strategy.”
Brad Bogolea · co-founder and CEO, Simbe Robotics
Inc., July 2025 · inc.com
He built that memory the slow way, a conversation at a time.
The guided interview, seven sections
About twenty-five minutes of questions sharp enough to do the thinking for you.
Your complete context pack, all four files
About Me, About My Work, Who Your Clone Is and the Setup Guide. Markdown for your AI, PDF for you.
Bonus one: The Letter From Next Year
Your clone writes to you from twelve months out, and it isn’t allowed to write the flattering version.
Bonus two: Your First Week
A seven-day PDF course that turns a download into a habit.
Privacy by design
No account, no upload. We couldn’t read your answers even if we wanted to.
Working this out cost me over £1,200 in courses and a year of evenings.
It costs you £29 and half an hour.
£29
one-time · about a month of the AI subscription you already pay for, paid once and kept forever
Instant delivery through Whop · Works with Claude and ChatGPT
Questions, answered straight
The basics
A guided interview and the files it produces. You answer questions for about twenty-five minutes; Second Self turns your answers into four short documents that describe who you are, what you're working toward, what's in your way, and how you want to be spoken to. You paste those into a project inside Claude or ChatGPT, and from then on every conversation starts with an AI that already knows all of it.
There's no app to keep, no login, and no monthly fee. What you're buying is the twenty-five minutes being good: the questions, the structure, and the setup that turns rough honest answers into something an AI can actually use.
Honestly? Yes, in the way you could cut your own hair. Everything Second Self produces is plain text you could theoretically have written. What you're paying for is the interview: twenty-five minutes of questions sharp enough that the answers organise themselves, structured into files your AI can actually use, plus the job-description file and setup routine most people never think to build. If you already keep a written brief about yourself pinned into your AI and update it monthly, save your money. You've built this by hand.
Yes, and if anything the personal side is the half that surprises people. The work section is entirely optional; plenty of people skip it and get a clone that's brilliant on the rest of life: decisions, planning around a real week, writing difficult messages, thinking something through at eleven at night.
You don't need a business, a following, a side hustle or a job title. You need a life with things in it you're trying to get done, which is everybody.
Most of those build a public clone of you, for an audience of fans, clients or followers, hosted on someone else's servers and charged for monthly. Second Self builds a private clone for you: an AI that knows you, working inside the subscription you already have. One-time payment, nothing hosted, nobody else ever talks to it.
Will it work for me?
If you can answer honest questions and copy-and-paste, you're qualified. There's no coding, no settings-diving, and the Setup Guide is written in plain English with the exact clicks for both Claude and ChatGPT.
The hard part of this isn't technical at all. It's being honest in the interview. That's the only skill it asks for.
Claude and ChatGPT get step-by-step setup instructions in the guide. The pack itself is plain text, and the shape (a project, standing instructions, your files) works anywhere that supports it, including Gemini and Copilot with light adaptation.
No. The interview and your files work regardless. Free plans are enough to run a clone; where a free plan lacks a feature like Projects, the Setup Guide shows the fallback route using custom instructions. Paid plans (around £18–£20 a month) simply give the clone more room and a better model.
About thirty minutes start to finish: roughly twenty-five for the interview and five to install. Most people do it inside a lunch break. You can also stop halfway; your draft stays saved on your device and waits for you.
It sometimes will. That's true of every AI tool, and no context file fixes it. AI can state something confidently and be flatly wrong, and it can quietly invent a specific detail you never gave it.
What context does change is the failure mode. An AI that knows your constraints stops producing advice that's irrelevant to your life; it doesn't stop producing the occasional confident error. So treat your clone as a very well-briefed colleague rather than an oracle: brilliant for drafting, thinking and pressure-testing, and always worth a read-through before anything goes out with your name on it. Anything with legal, medical or financial consequences still wants a professional.
Your privacy
Nowhere. The interview runs entirely in your browser. There's no account, no upload, and no Second Self database for your answers to sit in. The files exist in two places only: your device, and whichever AI project you paste them into. We never see them, by design.
That's between you and them, and it's worth two minutes of your time to check rather than take anyone's word for it, including ours. Both Claude and ChatGPT have a privacy or data-controls setting governing whether your conversations can be used to improve their models, and both let you change it. The Setup Guide points you at exactly where that setting lives in each, and suggests you look at it before you paste anything in.
What we can say without qualification is our side: Second Self never receives your answers at all, so there is nothing on our end to train on, sell, leak or hand over.
The rule the templates use: give the shape, not the specifics. Your AI needs to understand your situation, not hold your credentials.
Fine: your goals, your worries, how you make decisions, your working hours, roughly what you earn or charge as a range, the kind of clients you serve.
Generalise: money as a band rather than exact figures, clients as types rather than names, pricing philosophy rather than a live rate card.
Leave out entirely: passwords and account numbers, other people's private information, anything under NDA, and live legal matters. The interview is built to steer you this way, and every file stays editable, so if a line makes you uneasy, delete it.
Living with it
Barely. Your files are a few pages of plain text, and both Claude and ChatGPT are built to hold exactly this kind of standing context. Claude's Projects pull in only the relevant parts of a file rather than re-reading everything each time.
The practical guidance in the Setup Guide is the same one the AI companies give: keep the standing instructions short, keep the files focused, and clear out anything you've stopped using. Done that way, a context pack costs you almost nothing and saves you the paragraph of background you'd otherwise retype every session.
Every month or two, and it takes about ten minutes. This matters more than it sounds: a clone working from last spring's goals will give you last spring's advice, confidently.
Your answers stay editable on your device, so rebuilding the files is one click. There's also a shortcut the product teaches. Ask your clone "based on what I've told you this month, what in my About Me file is now out of date?" and it will do most of the work for you. And you can start again from scratch whenever life changes enough to deserve it.
Very little. The shape never changes (a project, standing instructions, your files) even when the buttons move, and the Setup Guide explains the shape as well as the clicks so you can find your way round a new menu.
The deeper reassurance is that your clone doesn't depend on me at all. It lives inside your own ChatGPT or Claude account, built from files you hold. If Second Self vanished tomorrow, your clone would carry on working exactly as it does today.
No. Second Self is a setup tool: it makes your AI dramatically better informed about you, and the two bonus exercises are reflective, not clinical. Your clone will only ever be as wise as the AI behind it plus the honesty of your files, and it's no substitute for professional support where that's what's needed.
The money
No, and no. One payment, once. There's no membership, no renewal, no card kept on file, and nothing to cancel, which makes Second Self unusual in a category where almost everything is monthly.
The only ongoing cost in this whole picture is the AI subscription you either already pay for or don't need. We don't take a cut of that either.
Thirty days, one message, full refund. No forms and no phone call. I'll process it the same day I read your message, and Whop returns it to your card, usually within five to ten working days depending on your bank. You keep the files either way.
The one thing I will do is ask what was missing. Not to talk you out of it: this is a new product and the people it didn't work for teach me more than the people it did. Answer in a line, or ignore the question entirely. It makes no difference to your refund, which is already on its way by the time I ask.
Your AI already has the intelligence.
Give it your story and context, and see what it can really do!
One-time · Both bonuses included · 30-day no-quibble guarantee
P.S. If you scrolled straight here (I always do): Second Self is a 30-minute guided setup that turns what you know about yourself into files that make ChatGPT or Claude actually know you. £29 once, two bonuses, 30-day no-quibble refund, and your answers never leave your browser.
The small print, in full
All of it on this page, none of it hidden behind a link. Open whichever you need.
Second Self is sold by Mark Speer, trading as Simple Soul, a sole trader based in the United Kingdom.
Write to hello@simple-soul.com and you will get a reply from me, not a helpdesk. If you have bought, messaging me through Whop reaches me just as well and keeps your purchase attached to the conversation.
I answer within two working days, usually sooner.
What I hold about you. Your email address and a record of your purchase. That is it.
What I never hold. Your interview answers. The interview runs entirely in your browser. Nothing you type is transmitted, stored or seen by me or anyone else. If you close the tab without saving your files they are gone, and I cannot recover them, because I never had them.
Who processes the payment. Whop acts as merchant of record. They take your payment details and handle the tax. I never see your card.
What I use your email for. Sending your purchase, answering you if you write, and occasional notes about Second Self. One click unsubscribes you from the last of those.
Your rights. Ask me for a copy of what I hold, or ask me to delete it, and I will do it within a month.
Who is responsible, and how to complain. For your email address and purchase record, the data controller is Mark Speer, trading as Simple Soul, in the United Kingdom. Whop handles your payment details under its own privacy policy. Write to me first at hello@simple-soul.com. If I don't put it right, you can complain to the Information Commissioner's Office at ico.org.uk.
Who you are buying from. Mark Speer, trading as Simple Soul, a sole trader in the United Kingdom. Payment is taken by Whop as merchant of record.
What you are buying. A one-time licence to use Second Self, comprising the guided interview, the context pack it generates from your answers, and the two bonuses. There is no subscription, no renewal and nothing to cancel.
What you may do with it. Use it yourself, as many times as you like, forever. The files it generates are yours outright. What you may not do is resell, republish or redistribute the interview or the bonuses, or pass them on as your own.
Delivery. Immediately on purchase, through Whop.
Your 14-day cancellation right. UK digital content normally carries a statutory 14-day right to cancel. Because Second Self is delivered immediately, you are asked at checkout to agree to immediate access and to acknowledge that the statutory right ends once access begins. This does not affect the 30-day refund below, which is more generous than the law requires.
What is not promised. Second Self makes your AI better informed about you. It cannot make an AI accurate. AI tools state things confidently and are sometimes wrong, and no context file changes that, so treat your clone as a very well-briefed colleague rather than an oracle and check anything with legal, medical or financial consequences with a professional.
Liability. Nothing here limits liability for death, personal injury or fraud. Otherwise, and to the extent the law allows, liability is limited to the amount you paid.
Law. These terms are governed by the law of England and Wales, and your statutory rights as a consumer are unaffected by anything on this page.
30-day no-quibble refund. If Second Self is not for you, message me through Whop within 30 days and say so. You do not have to give a reason. I will refund in full, and you keep the bonus PDFs.
I may ask what was missing so I can improve the product. Answering is entirely optional and makes no difference to your refund, which is already on its way by the time I ask.
Refunds are issued through Whop and normally reach your card within five to ten working days, depending on your bank.