gen-z-developers.sgit.ai → Synthetic readers → v2.0.0-rc2 → Interviews

The interviews, question by question

The same questions for every reader, so the answers read down a column as well as across a run.

1. In one sentence, what is this book?

Aisha Bello: It is a book that wants me to start a GenAI company by the last page, and so far it tells me why, how to price and sell, and how to build with no server, but not how to form the company.
Ellie Marsh: It is a book about how people who can describe what they want, not only people who code, can start a company with AI agents, written by someone who has built a lot of it himself.
Nadia Rahman: It is a book that tells an expert, in the author's own voice, that she can now build the first version of her company with agents, and mixes that with the story of how the author built and failed at his own products.
Martin Kowalski: A book for founders in the age of coding agents, in the author's voice, with the product-selling, process and tutorial material of a newsroom wrapped round a genuinely good middle on funding, shipping and agent control.
Catherine Doyle: A practical, personal argument that people with expertise in a field, and the generation entering work, can now build and start a GenAI company themselves, with agents, without servers, written in the first person by a security and software practitioner.

2. Who is it for? Is it for you, and which part?

Aisha Bello: It is for people starting a company with agents, and the author says Gen Z and Gen X, but the examples are mostly experts in their forties and big company buyers. I am the Gen Z reader he wants, but only parts 6, 7 and 9, and chapters 18, 19 and 40, were for me.
Ellie Marsh: It says it is for Gen Z and for their parents who know a field well. I am the first of those and I was glad to see non-coders named. The parts that were for me were chapters 4 (where I land, I think I am a villager), 8 (anyone who can brief an agent is a developer) and 13 (the Contacts folder). The back half is for someone who already builds.
Nadia Rahman: It is for me, a Generation X subject expert, and for my daughter. Chapter 1 describes me nearly exactly. The parts that are for me are 6 to 9, 13, and 17 to 21. Chapters 3 to 5, 10 to 12 and 23 to 27 are for someone else, a developer or a graduate.
Martin Kowalski: For the Gen Z graduate and for me, a Gen X expert. The parts for me are ch 17 on expert founders, ch 21 on funding, ch 24 and 25 on keys and agent control, and ch 27 on shipped. The job-hunting part (ch 10 to 12) and the storyboard and newsroom material (ch 34, 36, 41) are for someone else.
Catherine Doyle: It is for the founder in their forties or fifties with twenty years in a problem, and for the Gen Z reader entering work. The Gen X half is for me: chapters 1, 8, 9, 15, 17 to 21 are the ones I could sell. The Gen Z job-hunting chapters 10 to 12 sit awkwardly next to a founder book.

3. Did you believe it? What made you believe it, or not?

Aisha Bello: Partly. The honesty helps: the notes saying prices are hypotheses (ch 20), the GDPR correction (ch 24) and placeholders that say they are placeholders (ch 22, 37). What hurts is that a lot of the proof is the author's own companies, RiskMandate and SG/Send, which I cannot check.
Ellie Marsh: Mostly yes, which surprised me. I believed the spreadsheet stories in chapter 8 and the honest notes like 'I cannot now say who she was, so I will not put a name to it' in chapter 2. I believed less when the proof was always the author's own tools, like RiskMandate in chapters 19 and 39 and sgit in chapter 24, with no customers like me.
Nadia Rahman: Mostly yes, because he shows his failures: the thirty customers corrected to ten in chapter 18, the list of what the agent corrected in chapter 6, the unfinished chapters. I did not believe the opening of chapter 2, which rests on an essay he could not read, and I did not believe the code-review company at the end of chapter 17 had anything to do with me.
Martin Kowalski: Mostly, because of the self-correction. Ch 16 admits 30 customers was ten and nine months was eight. Ch 24 corrects his own GDPR claim. Ch 21 lays out the evidence against his own thesis. What stopped me believing it fully: three of the chapters I liked are about his own products, and the ch 17 founder examples are people he helps.
Catherine Doyle: Mostly, yes, in the argument and the numbers that are shown with their sources (chapters 14 and 19 especially). I believed it less where it relies on the author's own products as proof (chapters 24, 25, 39) and where it cannot read its own source (chapter 2's paywall). The honest hedging made me trust the author, and the 'For Dinis' notes made me trust the process while worrying about the text.

4. Where did it lose you, or make you guess?

Aisha Bello: Chapter 1 and 2 with the 2018 history, chapter 3 with graphs, and chapter 4 with Wardley maps, because I did not see why I needed them to decide job or company. Chapter 17's code review company lost me too. RiskMandate was never defined where I met it in 19.
Ellie Marsh: Chapter 3 (code as a graph, blast radius, sgit commands) made me think the book assumed I code. Chapter 6 lost me in the corrections list. In chapter 19 I had to guess what 'an Agent Behaviour Policy as an encrypted vault' actually is. Chapter 40 lost me for good with Worker, R2 bucket and Durable Object.
Nadia Rahman: In chapter 3, with the code section, in chapter 4, with the eight-row table, and in chapter 5, which I read as the author thinking aloud. I guessed what the harness is in chapter 7, and what a Durable Object is in chapter 40. And I was lost when chapter 17 turned into a business plan for a code-review company.
Martin Kowalski: Ch 2 lost me, it is mood. Ch 24 lost me in the key types. Ch 13 and ch 30 are the same chapter, and I stopped trusting the structure when I saw that. Ch 22 and 37 are placeholders for the very thing the title promises, forming a company.
Catherine Doyle: Chapter 3, on code as a fractal graph, lost me because I could not see why a founder needs it. Chapter 6 made me guess which picture the book relies on. Chapters 22 and 37 lost me by delivering nothing on the title.

5. The pacing: where did it drag, and where did it rush?

Aisha Bello: It drags in part 1 (ch 1 to 5) and in the middle chapters that retell the 2018 book first. It rushes where it matters most: ch 22 and 37 on forming a company are a few hundred words and say they are empty. Chapter 39 repeats 19, so it also drags there.
Ellie Marsh: It dragged in chapters 2, 5 and 9, which say the same hopeful thing several times, and again where chapters 13, 30, 19 and 39 repeat each other. It rushed through chapters 6 and 23, which are lists of terms and products, and chapter 40 is a rush of unexplained names.
Nadia Rahman: It dragged in chapters 2 to 5, about 8,000 words of wonder, graphs and maps before I built anything, and again in chapter 21 on funding. It rushed in chapter 22, which is a placeholder, and at the end in 37, 38 and 40, where vocabulary arrives with no time to define it.
Martin Kowalski: It dragged in ch 2, 5 and 41, and in the hands-on part where ch 32, 33 and 39 repeat arguments already made in ch 7 and 19. It rushed the company itself, ch 22 and 37, where the promise is greatest and the text is smallest.
Catherine Doyle: It drags in chapters 2, 3 and 5 (opening and philosophy), and in Part 9 where chapters 32, 33 and 39 repeat earlier material. It rushes chapters 22, 35 and 37, which are stubs or one-page promises.

6. The structure: does the order of the parts make sense? What would you move, cut or merge?

Aisha Bello: I would move part 9 (the hands-on chapters) to just after part 1, or interleave each hands-on chapter with the part it belongs to. I would merge 19 and 39, and 22 and 37, cut chapter 5 down to the pivot story, and move 8 and 28 to an appendix. I would put the job-or-company question in chapter 1.
Ellie Marsh: Part 9 should come much earlier or be threaded in next to the idea it practises, because the practical chapters (13, 30, 39) are where I woke up. I would cut the duplicated chapter 30 or 13, merge 19 and 39, move chapter 41 to the end as an afterword, and shrink the cloud and security chapters (23 to 25) to what a non-coder needs. The three placeholders (22, 37, 38) cannot be sold as chapters.
Nadia Rahman: I would move chapter 13 or the workspace afternoon up, to just after chapter 1 or 8, so a reader can do something in the first fifty pages. I would merge 13 and 30, 7 and 33, and cut or move 10 to 12 to a separate job-seeker book. Chapters 3 to 5 could be one. The hands-on part should not be at the end.
Martin Kowalski: Merge 13 and 30, 19 and 39, 12 and 36, 7 and 32 and 33. Cut 34 from a founder book. Move ch 41 out to the website, keep one page. The hands-on part would be better folded into the chapter each exercise belongs to than set as an eleven chapter appendix that repeats them.
Catherine Doyle: The order of the parts is sound until the end: frame, jobs, agents, judgement, company, serverless, long game, hands-on, process. I would merge chapters 13 and 30, merge 32 into 7, cut or move chapter 34 and 36 to a companion web page, move the process chapter 41 into an appendix, and bring Start the company forward to follow chapter 9 so the objective arrives sooner.

7. What is strongest? What would you underline, or send to somebody?

Aisha Bello: The test in chapter 18 and 19, give it away then take it away and see if they miss it, plus the Stripe fee arithmetic (below about £5 the fixed fee is the price). I would send those two to my course friend. Chapter 38's credits table too.
Ellie Marsh: Chapter 8: 'The language you already speak is the one you program in.' I would send chapter 8 and chapter 13 to my sister. I would underline the shipping definition in chapter 18: 'Not to plan, not to pitch, not to architect. Ship.' And chapter 19's card-fee arithmetic is true and useful.
Nadia Rahman: Chapter 8 on the briefing-only developer, chapter 9 on the prototype as a brief that is not yet a product, chapter 18 on the miss-it test, and chapter 20, the one who signs: dry, fair. I would send chapter 9 to my CIO and chapter 15's one-way and two-way doors to my risk committee.
Martin Kowalski: Ch 21 on funding by selling shares and lending back is the idea I would send round. The reach, mandate and gap model in ch 25 I would use tomorrow. The idle-cost test for serverless in ch 23 is a good one-liner. The Nokia inertia argument in ch 17.
Catherine Doyle: Chapter 15 on direction, chapter 19 on pricing with the Stripe arithmetic, chapter 9's renaming of vibe coding as explorer coding, and chapter 14's 'every mistake added a rule'. I would send chapter 19 to a founder I know.

8. What is weakest, or does not make sense?

Aisha Bello: The company chapters are empty, which is the book's whole promise. The history and the author's own products take up a lot of room. Chapter 29 on Portugal and chapter 34 on storyboards are about the book, not me.
Ellie Marsh: Chapter 40, the weekend chapter, which should be the proof and is a developer's recipe with stories of a failed product. Chapters 22 and 37 are empty on the thing the cover promises, a company. And the book keeps using the author's products as the examples, not a normal small business.
Nadia Rahman: The weakest are chapters 22 and 37, which are the promise, and the proportion of examples that are the author's own product, RiskMandate, SG/Send and sgit. Also the notes to Dinis that are left in the page. It sometimes reads as the author's portfolio rather than my route.
Martin Kowalski: The weakest are the placeholders (22, 35, 37), the duplicated workflow (13 and 30), and the notes to Dinis in the manuscript. Also the strong form of 'by the end of the book you have started a company' with no chapter that does it.
Catherine Doyle: Chapters 22 and 37, the placeholders, followed by chapter 41's process explanation and the repeated recipes. Also the notes to Dinis sitting in the text.

9. Did it read as written by a person, or by an AI? What made you think so?

Aisha Bello: A person, mostly. It is in the first person, tells stories with dates and has odd bits like the XZ aside and admitting stalled Leanpub books. But the notes to Dinis, the heavy citation of numbers and the repeated chapters feel like an agent assembled it.
Ellie Marsh: Mostly a person. It has odd, specific, human asides (the Spectrum, the blown-up phone line, the paywalled essay) and admits what it does not know. The AI side shows in the repeated structures, near-copied chapters (13 and 30, 19 and 39), the 'For Dinis' notes and the dense lists of facts with dates.
Nadia Rahman: Mostly a person, with a sharp voice, quirks like padel and the iMac, and some honest failure. But it also read as machine-made in places: the long lists of quoted figures, the repeating structure of 'In 2018 I wrote' at the start of nearly every chapter, and the same paragraphs in chapters 13 and 30, 7 and 33. The book says it is co-written, which I respect.
Martin Kowalski: Parts of it read as a person: the tangents, the admission his books stall, the 2010 keynote that half came true. Parts read as an agent working from sources: the repeated 'In 2018 I wrote' openings, the habit of closing each chapter on a neat line, and the repetition of the same phrases ('Execution just got cheap', 'twenty years inside a problem'). He tells you it is co-written, which I respect, and it still reads about 60 percent person.
Catherine Doyle: Both. The voice is clearly a person's: stories, opinions, mistakes. But the build method is machine-shaped: 'By the end of this chapter you will have' repeated in Part 9, uniform chapter lengths around 2,200 words, and statistics piled up in a pattern. The author says so himself in chapter 41, which is the right thing, but late.

10. Will you buy it? At what price, in which format? If not, what would have to change?

Aisha Bello: No, not at this version. If the company chapters were real I would pay £9.99 for the ebook, maybe, and read the hands-on part on my phone. Paperback at £19.99 only if a friend says it works. It would have to say how to form a company and drop the repeats.
Ellie Marsh: As it stands, maybe. At £9.99 on my Kindle, not £19.99 for the paperback. I would buy it if the practical chapters came first and I could see one non-coder example from start to finish, and not before the company chapters are written.
Nadia Rahman: Not yet. At £19.99 for the paperback if chapters 22 and 37 are written, and the personal-risk question of leaving a job gets an answer. I would not pay £9.99 for the ebook of the current text. To change my mind: the steps to form a company in the UK, and what it costs a person with a mortgage to do this.
Martin Kowalski: No, not at £9.99 in this form. At that price I would expect an edited book. If it were cut to 50,000 words, the placeholders written, the two overlapping chapters merged and the product examples diversified, I would pay £9.99 for the ebook and read it on my tablet.
Catherine Doyle: Not as it stands. I would pay £9.99 for the ebook and £19.99 for the paperback once the company chapters exist. If the first edition is free and this one is free to read online too, the paperback is a gift or a reference, and I would price it at £16.99 to £19.99.

11. One change. What is it?

Aisha Bello: Fill the company-forming chapters, and open with the job-or-company question and a first afternoon of action instead of 2018.
Ellie Marsh: Put a plain-language start-to-finish example early: someone who has never coded, with an idea like mine, going from idea to a first customer, in a weekend.
Nadia Rahman: Write chapters 22 and 37, and put the finished workspace afternoon before chapter 3. If I could only change one thing it would be the company chapters.
Martin Kowalski: Write the company chapters (22 and 37) from real steps, and cut the hands-on part to the chapters that do not repeat the argument.
Catherine Doyle: Write the company chapters. The promise of the book is that by the end you have started a company, and the UK, Lithuania, Portugal and EU chapters do not exist.

12. Would you give it to somebody else? Who?

Aisha Bello: Maybe a course friend who is also asking about starting something, and my mum, who has been inside her field for twenty years, for part 6 and chapter 17.
Ellie Marsh: Yes, chapters 8 and 13 as a short extract, to my sister and to a colleague who also has an idea. I would not give her the whole book yet.
Nadia Rahman: Yes, to two people. My daughter, for chapters 1, 2 and 8, because she keeps telling me to just build it. And my CIO, for chapter 9. I would not give it to my team as it is.
Martin Kowalski: Yes, two people: a Gen X friend who left a security vendor and is thinking about starting up, to read ch 21 and 25 only, and a former colleague who runs a small agent team, for ch 14 and 27. Not the whole book.
Catherine Doyle: Yes: to a founder who has twenty years in a regulated field and has said 'I wish someone would build this'. Not to a Gen Z reader first; for them the job chapters feel like a different book.

Publisher 1. Would you commission it, under your imprint? Yes, yes with conditions, or no.

Catherine Doyle: Yes with conditions: the missing company chapters written, a clear single audience, and a cut of 20 to 25 percent of repeated and process material.

Publisher 2. Who buys it, and where does it sit in a bookshop or a store?

Catherine Doyle: Business and technology shelf, founders and startups, next to books about AI for entrepreneurs. Online it sells under AI for founders, and as a gift to a family member who has an idea. Corporate bulk orders would suit the agent-control chapters.

Publisher 3. What are the risks: the market, the dating of the material, the free first edition, the author's own products, advice that reads as legal or tax advice, the AI co-writing?

Catherine Doyle: Market: a crowded AI shelf where two of my own did differently. Dating: the programme tables, price and statistics. Free first edition and a free online edition. The author's own products (sgit, SG/Send, RiskMandate) throughout. Advice that reads as legal or tax advice in chapters 21 and 24, despite the disclaimers. The AI co-writing, which some bookshops will notice. And the 'second edition or new book' question remains undecided.

Publisher 4. What would you change before you signed: title, length, audience, structure?

Catherine Doyle: Title: I would drop 'Generation Z Developers' from the front and sell 'Gen ZX' or a plain 'Start a GenAI Company'. Length: cut towards 60,000 words by merging the duplicated chapters. Audience: lead with the expert founder. Structure: move the process to an appendix and bring the company part forward.

Publisher 5. What must the author bring: platform, endorsements, a date?

Catherine Doyle: A platform: a mailing list, a talk circuit, a named set of founders who used it. Endorsements from two or three founders and an investor who has read it. A date: a finished manuscript with the company chapters written, and a plan for refreshing the programme tables each year.