Home / Articles / Connect your dots backwards: a look-back interview with your LLM, for founders, to find the non-linear path, the doors that were not what they seemed, and where a startup may be hiding

Connect your dots backwards: a look-back interview with your LLM, for founders, to find the non-linear path, the doors that were not what they seemed, and where a startup may be hiding

By Dinis Cruz · 2026-10-21 · article v1.0.0 · newsroom v0.13.21 · gen-z-bookfoundersstartupslearningtoolkitarticle

Abstract: The book has a chapter called Connect the dots backwards, after Steve Jobs's line that you can only connect the dots looking backwards. I think founders should do it for themselves, now, and that it is easy to underestimate how much it gives. So here is a proposal, and a small toolkit: get your LLM to interview you about your life, your experience, your career and what you learned, and see whether you can already connect your dots backwards in ways that surprise you. Most people's paths are not linear, mine included. We mostly see the success stories, often retold with the challenges and the silly-sounding experiments glossed over, yet those experiments are often what made the person. That is why, when I interviewed people, I always looked at what else they had done. The interview finds the side roads, the doors that felt like one-way doors and were not, and the ones that felt small and set the path. It asks what you can evidence, because memory rewrites itself a little every time we recall it. It is an answer to impostor syndrome made of facts, not reassurance. And it may show you the startup hiding in something you were passionate about before, where it did not work out at the time.

In short

• Connect your dots backwards, now. Not at the end of a career: as a founder, today.
• Let your LLM interview you. Your life, your work, what you learned, one question at a time.
• Most paths are not linear. And the success stories we see were often tidied up.
• The side roads matter most. The experiments that sound silly in hindsight often made the person.
• Look at the doors. Which felt like one-way doors and were not, and the other way round.
• Check what memory rewrote. Every recollection changes the memory a little.
• Find where a startup may be hiding. In an old passion where it did not work out at the time.

Why do it now

The book has a chapter called Connect the dots backwards, after Steve Jobs's commencement address at Stanford on 12 June 2005: "you can't connect the dots looking forward; you can only connect them looking backwards." His example was a calligraphy class he took after dropping out, which came back ten years later as the typography of the first Macintosh.

The chapter tells some of my own dots. What I want to propose now is that founders do this for themselves, deliberately, and soon. I think it is easy to underestimate. Looking back at your own path, honestly and with help, shows you things about yourself that you cannot see while you are walking it.

Most paths are not linear

Most people's paths are not linear, and especially for my generation, they have not been. I went from the ZX Spectrum to security, from security to boards, from boards to news, and from news to vaults; none of it was planned that way. As I wrote in not a straight line, progress does not go in a straight line, and the thinking is never wasted.

The problem is that we mostly see the success stories. And we see them retold, sometimes by the people themselves, sometimes by their industry, with the challenges and the problems along the way glossed over. That gives the next generation a false picture: that a path is supposed to be a straight line, and that if yours is not, something is wrong.

The experiments are often the opposite of what the retelling suggests. Some of them sound silly in hindsight, and sometimes they are exactly what made that person. That is why, when I was interviewing people for a job, I always looked for the other things they had done. They were often more interesting than the job they ended up applying for, and they said more about how the person thinks. That multidisciplinary side, as the chapter says, is more and more where the edge is.

Let your LLM interview you

So here is the exercise: get your LLM to interview you. About your life, your experience, what you learned, your career and the paths you took. Not a CV: a conversation, one question at a time, in your words, with an interviewer that keeps asking "and what else were you doing then?"

The interview has seven stages:

• The timeline. Every school, job, city and project, in order, with rough dates.
• The side roads. What you did that was not the job: side projects, hobbies, things you taught yourself, communities, the startup that did not work.
• The doors. Each turn in the path, how it felt at the time, and what it really was.
• What each dot taught you. A skill, a person, an idea, in one line each.
• The connections. Links the interviewer proposes between dots, which you confirm or reject; the surprising ones are marked.
• Remembered or evidenced. What you can show, and what may have been rewritten.
• Where a startup may be hiding. Old passions, unsolved problems, jobs you know from the inside.

The doors

Jeff Bezos, in his letter to shareholders for 2015, split decisions into two kinds. One-way doors: "consequential and irreversible or nearly irreversible", where "you can't get back to where you were before." And two-way doors: "changeable, reversible", where "You can reopen the door and go back through." The newsroom uses that split in how cheap is it to undo?, and the book uses it for walking a path you cannot see.

Looking back, the interesting doors are the ones we read wrong at the time. The two-way doors that felt like one-way doors: the job you were afraid to leave, the course you were afraid to drop, where fear kept you there longer than it had to. And the one-way doors that felt like two-way doors: the side project, the meetup, the email you nearly did not send, which set the path and nobody noticed. Knowing which of yours were which is how you get better at reading the next one.

Memory rewrites itself

There is a reason to ask for evidence. Our brains do a good job of rewriting history a little, because that is how memory works. My understanding is that every time we recall a memory, we change it a little and store that recollection. Research supports a version of this: in 2000, Karim Nader, Glenn Schafe and Joseph LeDoux showed in Nature that consolidated fear memories, when recalled, return to a state where they have to be stored again, a process now called reconsolidation. The science is narrower than the everyday version, but the lesson for this exercise holds: what you remember is not a recording.

So the interview asks, for each dot, what you can show: an old email, a photo, a post, a repository, a CV from the time, a commit history. Each dot is marked as remembered or evidenced. The evidenced ones are the dots you can stand on.

Impostor syndrome, answered with facts

A lot of people have impostor syndrome. The name comes from a 1978 paper by Pauline Rose Clance and Suzanne Imes, on "the imposter phenomenon", an internal sense of intellectual phoniness in people who are, by any outside measure, successful. And part of what feeds it is that same rewriting: in our own heads, we remember the failures sharply and the progress vaguely.

The look-back interview is an answer made of facts, not reassurance. When the timeline, the side roads and the evidence are written down, the path is there to read: what you did, what it taught you, and how it connects to what you are doing now. That is harder to argue with than a pep talk, including when the one arguing is you.

Where a startup may be hiding

The last stage is the practical one for a founder. Look at the side roads again. The things you were passionate about, where it did not work out at the time: the technology was not there, the money was not there, you were not ready, the market was not ready. Some of those are worth trying again now, especially if it is one of the jobs that never scaled and you know it from the inside.

That is where this connects to startups.sgit.ai, whose finder is an interview that helps a founder zoom in on what to ship, to whom, by when. The look-back interview is its companion: it finds the threads; the finder helps you pick one and ship it.

The prompt

This is the first version. Paste it into Claude, ChatGPT or any assistant that can hold a long conversation, and set aside an hour, or a few shorter sessions.

You are going to interview me about my path, so that I can connect my dots
backwards. You are not passing judgement, and you are not here to reassure me.
Your job is to help me see my path as it really was.

Run the interview in seven stages, one question at a time, in plain words.
After each answer, say in one line what you noted, so I can correct you.

1. The timeline. Every school, job, city and project, in order, with rough
   dates. Keep asking "and then?" until we reach today.
2. The side roads. For each period, ask what else I was doing that was not
   the job: side projects, hobbies, things I taught myself, communities,
   ventures that failed. Ask for the experiments that sound silly now.
3. The doors. For each turn in the path, ask how it felt at the time (a big,
   one-way decision, or a small, reversible one) and what it turned out to be.
   Mark the ones I read wrong.
4. What each dot taught me. One line each: a skill, a person, an idea.
5. The connections. Propose links between dots, where something from one
   was used in another. I confirm or reject each. Mark the ones that surprise me.
6. Remembered or evidenced. For each important dot, ask what I can show: an
   email, a photo, a post, a repository, a CV from the time. Mark each dot
   "remembered" or "evidenced". Do not argue with my memory; just ask.
7. Where a startup may be hiding. From the side roads and the connections,
   list the things I was passionate about where it did not work out at the
   time, and ask what has changed since.

Rules for you:
- Use my words. Where you summarise, say so, and let me rewrite it.
- Ask for failures and pauses as much as for successes.
- Do not flatter me, and do not tell me how I should feel.
- If I am vague, ask once more; if I am still vague, write down what I said.
- Do not invent dots, dates or connections. Everything comes from me.

At the end, write the record:
- my timeline, as a table;
- the side roads;
- the doors, with how each felt and what it was;
- the connections, with the surprising ones marked;
- each dot marked remembered or evidenced;
- three threads where a startup may be hiding;
- one honest paragraph about my path, in my words, that I could share with
  someone starting out.

Keep the record somewhere private that you control, such as your own vault. It is about you, and it is yours. The last paragraph, if you choose, is the one to share.

A mini toolkit, next

This is a prompt today. It should become a small toolkit, the same shape as the startups finder: the interview as data, with its stages and questions and the reason for each; the prompt; and a record template, so the result is comparable over time and can be redone in a year to see what moved. Its natural home is beside the finder on startups.sgit.ai, and that is the proposal for that site.

What this does not claim

• It is not therapy. It is a structured look back; if it brings up something hard, talk to a person.
• The LLM can be wrong about connections. You confirm or reject each one.
• Reconsolidation is narrower than the everyday version. The research is on recalled fear memories.
• The impostor phenomenon is a description, not a diagnosis. As its authors used it.
• The prompt is a first version. It has not been tested with founders yet.
• My own dots are summarised from memory in the book. As it says.

Where this comes from

A voice memo of 21 October 2026. The book's chapter Connect the dots backwards, and its quote from Steve Jobs's Stanford commencement address of 12 June 2005. Jeff Bezos, letter to shareholders for 2015. Karim Nader, Glenn Schafe and Joseph LeDoux, Fear memories require protein synthesis in the amygdala for reconsolidation after retrieval, Nature 406, 2000. Pauline Rose Clance and Suzanne Imes, The imposter phenomenon in high achieving women, 1978. startups.sgit.ai and its finder, read the same day. Related: the jobs that never scaled, programming, taught upside down, how cheap is it to undo?, not a straight line. Written up and checked by this site's agent, newsroom.writer; the editorial responsibility is mine.

Want the next issue by email. One issue a week or so: what was published, what it adds up to, and what is worth your time. Subscribe to the SGit Newsroom →

← All articles