Bravo! Burnout is all to real, and it can be devastating to a writer. The mental exhaustion and inability to write from your soul is as dehumanizing and demoralizing as anything an author can face. Finding a way out, whether it's AI or something else, and getting your creativity back is important.
AI can help writers in other ways, without doing the writing for them. While generic chats are not particularly good at giving feedback, there are ways to have AI actually provide valuable coaching most writers never have access to. Here's an example of a critique I had AI do on an old chapter sitting abandoned in an archive folder on a backup drive. This critique is from a custom app I'm putting together.
This feedback is something I can actually do something with, while keeping me in charge of the story.
------------
If this is indeed Chapter 3—occurring after the Berlin heist in Chapter 1—the structure of this scene is currently broken.
Why? Because the characters are acting as if Chapter 1 never happened.
A scene is a unit of change. It needs a Goal, a Conflict, and a Disaster/New Direction. Let’s break down where your current structure is failing the narrative flow and how to rebuild it.
1. The Setup (The Goal)
Current Structure: Natalie enters the café to meet a contact. She expects... we aren't sure. A job offer?
The Problem: If she just flew in from Berlin with a stolen, encrypted drive, her goal is specific and urgent: "Hand off the hot potato." She wants to give him the drive, get her pat on the back, and go home.
The Fix: Establish the physical weight of the drive immediately. She shouldn't be "waiting for him to speak." She should be sliding that drive across the table the moment she sits down.
2. The Interaction (The Conflict)
Current Structure: The Mentor explains the plot to her ("There's been a breach...").
The Problem: This is redundant exposition. The reader knows there was a breach; they watched Natalie commit it (or investigate it) in Chapter 1. Having a character explain the plot to the protagonist after she has already lived it kills the tension.
The Fix: Change the conflict.
Old Conflict: He wants to hire her; she is reluctant.
New Conflict (Stronger): She wants to log the evidence officially; he refuses to take it.
Dialogue shift: "I can't take that, Nat. And you can't log it. If that drive enters the system, the people who killed Henning will know where you are in ten minutes."
3. The Turn (The Twist)
Current Structure: He mentions "Viktor" and "recursive loops" to hook her.
The Problem: This is a "hook," but it’s too abstract. It feels like a briefing dossier.
The Fix: Make the turn personal and immediate.
Instead of telling her about Viktor, he should show her that she is already compromised.
Example: He slides a photo across the table. It’s a surveillance shot of her in Berlin, taken by someone other than the police.
The realization: "You aren't being recruited, Nat. You're being hunted. I'm the only one giving you a head start."
4. The Resolution (The Disaster)
Current Structure: She leaves, thinking about how she will accept the mission because of Viktor.
The Problem: This is a "soft" ending. It’s an internal decision.
The Fix: The scene needs to propel her into Chapter 4.
If she keeps the drive (because he won't take it), she is now a fugitive.
The scene shouldn't end with her thinking about accepting. It should end with her realizing she has no choice.
Closing Image: Not her walking into the cold thinking about trust. It should be her checking her phone and realizing her official credentials have been revoked. Or seeing the "government sedan" and realizing they are there to kill her, not watch her.
Summary of the Structural Re-Write
To make this work as Chapter 3, you need to invert the power dynamic.
Entrance: She comes in hot. She wants to dump the evidence and get out.
Rejection: The Mentor refuses the package. He shatters her expectation of safety.
Revelation: He reveals that the "official channels" are the enemy. He links the Berlin murder to the "Viktor" backstory.
The Push: He gives her a lead (a name, a location) but no support. She is on her own.
Exit: She leaves the café more in danger than when she entered.
The Golden Rule for Chapter 3: The reader is waiting for the consequence of the theft in Chapter 1. This scene must be that consequence. Do not explain the plot; complicate the plot.
I’m also finding AI very helpful for this type of development to see where the problem is and getting ideas on how to fix it. It sparks my imagination and I can run with it, rather than stay stuck.
I've been struggling with this and I'm glad you took the time to lay out your thoughts because they helped crystallize what I assumed were inherent contradictions into a framework that made sense to my overworked and addled brain. My path forward seems a bit straighter after reading your post.
I think I'd argue that this could go either way. I just read your article top to bottom, and I have no idea whether you used an AI to help you write it or not. In my eyes, that's a VERY positive thing. I've seen well over two million words of AI text at this point (might be over three million, come to think of it), between editing gigs, books I've read on Kindle, online articles, and so on. I generally consider myself pretty good at spotting unaltered AI output most of the time.
Although it's getting tougher. Someone did a flash fiction test last autumn, and I only got 3/4 correct. ;)
Anyway, my point is I read these articles because they clearly convey your voice. Whatever process you're using, I can still hear Steph coming through in those words. I enjoy getting *your* take on things. I like reading *your* thoughts on a topic. It's always interesting and enlightening.
On the flip side, I've seen a few other Substacks where the posts are generally just raw AI, or very close to raw AI. The person has clearly asked a chatbot "write me a blog post about X, hitting these three major topics," and then just posted the results. These essays tend to be bland, boring, and hit only the surface of a topic. It's like reading the Google search summary on something - they're not really deep-diving into the nuances or personal feelings about a topic.
If I wanted the answer to "does AI make writers more or less human in their output," I could just ask the AI myself.
If I wanted to see "what does Steph Pajonas, one of the world's leading experts on the intersection of AI and publishing, think about the human side of writing with AI?" Well, there's only two ways I can get that info. I can call her and ask, or I can read her blog post on the topic. ;) Because the blog post was carefully constructed in a manner that rings of voice, personal experience, and individual feelings, it's interesting and valuable.
But if the post was the raw AI answer that anyone could get just by asking ChatGPT, that's neither. And therein lies the key, in my experience.
Our voices are why readers follow us. Learning to produce work that resounds with our experiences, feelings, ideals, and emotions is how we build fans and followers, regardless what tools are being used to write the work.
Exactly! And if we have a weakness in our writing, like transitions from one thought to the next so it flows and doesn’t ask the reader to make cognitive leaps, using a tool like AI just helps my writing be better. It’s still my thoughts, my ideas, but now they just flow together more seamlessly.
I appreciate your thoughtful words and I think you've touched on key elements for how we can use AI to keep us moving and that we can use it as a thought partner and that doesn't diminish our creativity or lessen us as writers. In my day job when I am not writing I've met with numerous AI architects from major tech companies and they say the same thing.
This is exactly what I did, months ago. I fed it my full manuscript, and it threw out brilliant ideas about themes I had not fully realized (the depth) in my work. Now, I have 365 days' worth of facebook post ideas and 52 weeks' worth of substack post ideas. It's still a ton of work, and I may not be as thorough with my next book (probably will not), but it has been, and is continuing to be, educational and interesting for me.
Congrats. Super exciting. May I ask a question? You say AI responded with many ideas. My question: Did any of those topics touch on your predetermined theme(s)?
I just did something similar. I was writing my newsletter and wanted to include an excerpt of the book I’m in the polishing phase. I had a theme for my newsletter and asked AI if there was a scene in my book that would illustrate it. Within seconds it gave me a scene that perfectly fit. My book. My words… Instead of spending hours trying to find something but not seeing it (you know the can’t see the forest for the trees) and giving up, AI found a perfect match I could share with my readers.
My goal is to send weekly emails to my list. In the past, I’d do great for a month or two, then not know what to say or send and ghost my readers for another year. This year, I worked with AI to come up with a plan that fits me and what I write. It helped me see the threads I could weave throughout my emails so now instead of random thoughts, there’s a through line and story. It makes it easy for me to write the emails.
I love the line “Nobody shames a chef for using a food processor.” For the last two years my husband has been turning a used ambulance we bought into an RV. He is an experienced builder who loves learning new things, so he taught himself how to use a 3D modeling program and he designed the entire thing virtually. When it came time to actually build the 19 dovetailed drawers I convinced him to send the drawings out to a CNC machine. He had created the design, let the machine do the grunt work. Machines have been augmenting humans since the Industrial Revolution. Now, we have cognitive augmentation.
I love this post! I started using Claude and ChatGPT last summer. It was just an experiment, but one that helped me finish writing book 1 of my new mystery series, and book 2 draft complete. Chat never wrote them for me. But it did give suggestions that deepened my world in ways I would never have thought of on my own. And I highly doubt I would have completed either one without their input.
Upload the whole book. I have all my books saved in Markdown format so I can do this easily. All the frontier models have context windows to handle a good sized book plus more.
You’ve seen it and not realized what it is. It’s a lightweight formatting that indicates things like italics, bold, headings, lists, etc. https://www.markdownguide.org/getting-started/
Bravo! Burnout is all to real, and it can be devastating to a writer. The mental exhaustion and inability to write from your soul is as dehumanizing and demoralizing as anything an author can face. Finding a way out, whether it's AI or something else, and getting your creativity back is important.
AI can help writers in other ways, without doing the writing for them. While generic chats are not particularly good at giving feedback, there are ways to have AI actually provide valuable coaching most writers never have access to. Here's an example of a critique I had AI do on an old chapter sitting abandoned in an archive folder on a backup drive. This critique is from a custom app I'm putting together.
This feedback is something I can actually do something with, while keeping me in charge of the story.
------------
If this is indeed Chapter 3—occurring after the Berlin heist in Chapter 1—the structure of this scene is currently broken.
Why? Because the characters are acting as if Chapter 1 never happened.
A scene is a unit of change. It needs a Goal, a Conflict, and a Disaster/New Direction. Let’s break down where your current structure is failing the narrative flow and how to rebuild it.
1. The Setup (The Goal)
Current Structure: Natalie enters the café to meet a contact. She expects... we aren't sure. A job offer?
The Problem: If she just flew in from Berlin with a stolen, encrypted drive, her goal is specific and urgent: "Hand off the hot potato." She wants to give him the drive, get her pat on the back, and go home.
The Fix: Establish the physical weight of the drive immediately. She shouldn't be "waiting for him to speak." She should be sliding that drive across the table the moment she sits down.
2. The Interaction (The Conflict)
Current Structure: The Mentor explains the plot to her ("There's been a breach...").
The Problem: This is redundant exposition. The reader knows there was a breach; they watched Natalie commit it (or investigate it) in Chapter 1. Having a character explain the plot to the protagonist after she has already lived it kills the tension.
The Fix: Change the conflict.
Old Conflict: He wants to hire her; she is reluctant.
New Conflict (Stronger): She wants to log the evidence officially; he refuses to take it.
Dialogue shift: "I can't take that, Nat. And you can't log it. If that drive enters the system, the people who killed Henning will know where you are in ten minutes."
3. The Turn (The Twist)
Current Structure: He mentions "Viktor" and "recursive loops" to hook her.
The Problem: This is a "hook," but it’s too abstract. It feels like a briefing dossier.
The Fix: Make the turn personal and immediate.
Instead of telling her about Viktor, he should show her that she is already compromised.
Example: He slides a photo across the table. It’s a surveillance shot of her in Berlin, taken by someone other than the police.
The realization: "You aren't being recruited, Nat. You're being hunted. I'm the only one giving you a head start."
4. The Resolution (The Disaster)
Current Structure: She leaves, thinking about how she will accept the mission because of Viktor.
The Problem: This is a "soft" ending. It’s an internal decision.
The Fix: The scene needs to propel her into Chapter 4.
If she keeps the drive (because he won't take it), she is now a fugitive.
The scene shouldn't end with her thinking about accepting. It should end with her realizing she has no choice.
Closing Image: Not her walking into the cold thinking about trust. It should be her checking her phone and realizing her official credentials have been revoked. Or seeing the "government sedan" and realizing they are there to kill her, not watch her.
Summary of the Structural Re-Write
To make this work as Chapter 3, you need to invert the power dynamic.
Entrance: She comes in hot. She wants to dump the evidence and get out.
Rejection: The Mentor refuses the package. He shatters her expectation of safety.
Revelation: He reveals that the "official channels" are the enemy. He links the Berlin murder to the "Viktor" backstory.
The Push: He gives her a lead (a name, a location) but no support. She is on her own.
Exit: She leaves the café more in danger than when she entered.
The Golden Rule for Chapter 3: The reader is waiting for the consequence of the theft in Chapter 1. This scene must be that consequence. Do not explain the plot; complicate the plot.
I’m also finding AI very helpful for this type of development to see where the problem is and getting ideas on how to fix it. It sparks my imagination and I can run with it, rather than stay stuck.
I've been struggling with this and I'm glad you took the time to lay out your thoughts because they helped crystallize what I assumed were inherent contradictions into a framework that made sense to my overworked and addled brain. My path forward seems a bit straighter after reading your post.
I think I'd argue that this could go either way. I just read your article top to bottom, and I have no idea whether you used an AI to help you write it or not. In my eyes, that's a VERY positive thing. I've seen well over two million words of AI text at this point (might be over three million, come to think of it), between editing gigs, books I've read on Kindle, online articles, and so on. I generally consider myself pretty good at spotting unaltered AI output most of the time.
Although it's getting tougher. Someone did a flash fiction test last autumn, and I only got 3/4 correct. ;)
Anyway, my point is I read these articles because they clearly convey your voice. Whatever process you're using, I can still hear Steph coming through in those words. I enjoy getting *your* take on things. I like reading *your* thoughts on a topic. It's always interesting and enlightening.
On the flip side, I've seen a few other Substacks where the posts are generally just raw AI, or very close to raw AI. The person has clearly asked a chatbot "write me a blog post about X, hitting these three major topics," and then just posted the results. These essays tend to be bland, boring, and hit only the surface of a topic. It's like reading the Google search summary on something - they're not really deep-diving into the nuances or personal feelings about a topic.
If I wanted the answer to "does AI make writers more or less human in their output," I could just ask the AI myself.
If I wanted to see "what does Steph Pajonas, one of the world's leading experts on the intersection of AI and publishing, think about the human side of writing with AI?" Well, there's only two ways I can get that info. I can call her and ask, or I can read her blog post on the topic. ;) Because the blog post was carefully constructed in a manner that rings of voice, personal experience, and individual feelings, it's interesting and valuable.
But if the post was the raw AI answer that anyone could get just by asking ChatGPT, that's neither. And therein lies the key, in my experience.
Our voices are why readers follow us. Learning to produce work that resounds with our experiences, feelings, ideals, and emotions is how we build fans and followers, regardless what tools are being used to write the work.
Exactly! And if we have a weakness in our writing, like transitions from one thought to the next so it flows and doesn’t ask the reader to make cognitive leaps, using a tool like AI just helps my writing be better. It’s still my thoughts, my ideas, but now they just flow together more seamlessly.
I appreciate your thoughtful words and I think you've touched on key elements for how we can use AI to keep us moving and that we can use it as a thought partner and that doesn't diminish our creativity or lessen us as writers. In my day job when I am not writing I've met with numerous AI architects from major tech companies and they say the same thing.
This is exactly what I did, months ago. I fed it my full manuscript, and it threw out brilliant ideas about themes I had not fully realized (the depth) in my work. Now, I have 365 days' worth of facebook post ideas and 52 weeks' worth of substack post ideas. It's still a ton of work, and I may not be as thorough with my next book (probably will not), but it has been, and is continuing to be, educational and interesting for me.
Congrats. Super exciting. May I ask a question? You say AI responded with many ideas. My question: Did any of those topics touch on your predetermined theme(s)?
Oh yes. Sometimes it just highlights what I already knew about my story but was unable to articulate.
I just did something similar. I was writing my newsletter and wanted to include an excerpt of the book I’m in the polishing phase. I had a theme for my newsletter and asked AI if there was a scene in my book that would illustrate it. Within seconds it gave me a scene that perfectly fit. My book. My words… Instead of spending hours trying to find something but not seeing it (you know the can’t see the forest for the trees) and giving up, AI found a perfect match I could share with my readers.
My goal is to send weekly emails to my list. In the past, I’d do great for a month or two, then not know what to say or send and ghost my readers for another year. This year, I worked with AI to come up with a plan that fits me and what I write. It helped me see the threads I could weave throughout my emails so now instead of random thoughts, there’s a through line and story. It makes it easy for me to write the emails.
I love the line “Nobody shames a chef for using a food processor.” For the last two years my husband has been turning a used ambulance we bought into an RV. He is an experienced builder who loves learning new things, so he taught himself how to use a 3D modeling program and he designed the entire thing virtually. When it came time to actually build the 19 dovetailed drawers I convinced him to send the drawings out to a CNC machine. He had created the design, let the machine do the grunt work. Machines have been augmenting humans since the Industrial Revolution. Now, we have cognitive augmentation.
Love this!! :)
Great to hear. Wanted so much for you to confirm that part. But you are equally right, AI provides many ideas to work with. Thank you-- and good job!
I love this post! I started using Claude and ChatGPT last summer. It was just an experiment, but one that helped me finish writing book 1 of my new mystery series, and book 2 draft complete. Chat never wrote them for me. But it did give suggestions that deepened my world in ways I would never have thought of on my own. And I highly doubt I would have completed either one without their input.
So, when you say you "fed it your book", did you actually upload the file or cut and paste?
Upload the whole book. I have all my books saved in Markdown format so I can do this easily. All the frontier models have context windows to handle a good sized book plus more.
Learning something new here. What's Markdown format?
You’ve seen it and not realized what it is. It’s a lightweight formatting that indicates things like italics, bold, headings, lists, etc. https://www.markdownguide.org/getting-started/