It started with trolls.
Not the fairytale kind. The kind that sit behind a keyboard, buy a book on Amazon, read enough of it to feel smug, leave a one-star review that says no feedback at all... and then return it. Amazon refunds them immediately. Out of the writer's royalties.
The review stays up.
The algorithm sees the one-star feedback. Tanks the distribution. Kills the sales. The writer emails Amazon. Amazon does nothing. The writer posts about it in a writers' group, furious and helpless, and forty other writers pile on with their own horror stories, and everyone agrees it's broken and unfair and something should be done...
And nothing gets done.
So I started thinking about it. Not the trolls specifically, but the underlying problem. Writers need feedback. Real feedback. Not "your enemies are tanking your reviews" feedback and not "your mum thinks it's wonderful" feedback. Something in between that actually serves the craft.
I mean, think about it.
Your family reads your story. They love you. So they tell you it's brilliant. Maybe it is. Maybe it isn't. You can't tell, because you know they love you, so the feedback is contaminated before it arrives. You can't trust it. And if you believe it when you shouldn't... you walk out into the real world thinking you're a better writer than you are, and the real world will correct that impression in the most crushing way possible. Some people never write again after that correction.
Then there's the other extreme. Trolls. People online who are harsh for sport, not because they want you to improve. That's not feedback, that's just someone using your work as target practice.
So. Neither of those things helps a writer get better.
What Honest AI Feedback Looks Like
And then I looked at AI.
ChatGPT released a model... I don't know which one it was, but you could tell immediately. Someone worked it out pretty fast. You could pitch it the worst business idea you've ever heard, something that would fail spectacularly on day one, and it would tell you it was visionary. Groundbreaking. You're going to make millions. Seriously. That's not an exaggeration, that's what people were reporting. Shit on a stick, and the AI is telling you it's gold.
Come on.
That's not feedback. That's an AI equivalent of your mum. Sweet. Useless for growth.
So that became the first design challenge. How do you build something that won't do that?
Turns out it's harder than it sounds. I spent a year on it. A year of prompting, testing, refining, rebuilding. I put my own old stories in, stories I wrote in my twenties and thirties, stories I knew had problems, stories that were rushed, stories with errors I could still see. And every single time, the model would come back and tell me they were brilliant. Four out of five. Four and a half out of five.
Right.
There's no room in that scoring to distinguish between "writer who's getting there" and Edgar Allan Poe. There's no room for improvement if you're already sitting at four and a half. And I knew my old stories weren't four and a half. I knew that. But the AI couldn't tell me what I already knew was true.
That was the problem to solve.
How Feedback Lands Matters As Much As What It Says
Not just the scoring. The whole philosophy of how the feedback lands. Because a writer who gets a brutal review with no context, no structure, no "here's what's working before we talk about what isn't"... that writer shuts down. Doesn't matter if every word of the criticism is accurate. If it lands wrong, it doesn't land at all.
So I built it around a simple structure. What's working. What needs work. What's working. The first positive puts you in a frame of mind where you can actually hear the criticism. The criticism lands with room to breathe. The final positive sends you away with the feeling that there's something real here worth fixing. Not crushed. Motivated.
And I kept testing it on my own stories until one day it reviewed Becoming Forsaken and gave me a 2.75 out of 5.
I sat there for a second.
It stung a bit, yeah.
And then I thought... perfect. Fucking perfect.
It caught a character name swap I'd missed. It called out generic characterisation. It flagged dialogue as functional rather than alive. All things I could actually do something with. All things a developmental editor charges hundreds of dollars to tell you.
That's when I knew the logic was right.
I also found a flaw in my own build during testing. The tool was scoring a zero for Cultural Sensitivity on stories that simply didn't have any cultural elements requiring evaluation. Penalising a story for something that doesn't apply to it. So I fixed it. If a category doesn't apply to your story, it gets excluded from the score entirely. The diagnostic shifts from 20 points to however many actually matter for your work. Your score reflects what's actually on the page, nothing else.
That's when I knew the logic was right.
The SensAI Triad
SensAI: The AI that won't write your story, but will tell you the truth about it.
Now. I want to be really clear about something, because this is the part that matters most to me.
SensAI Triad tools dos not write for you.
I built it specifically so that it never would. It reads your work. It analyses the architecture of your story. It finds the cracks. And then it tells you where they are so that you can fix them. You. The writer. The human with the voice and the ideas and the years of craft.
There's a wave right now of AI tools that will generate your story, write your chapters, fill your plot holes, give you dialogue, basically hand you a finished manuscript while you sit there supervising. And I understand the appeal, I do. But those tools aren't for writers. They're for people who want a book with their name on it.
Writers write.
What writers actually need is what they've always needed. Someone who'll tell them the truth about the work without an agenda. Without protecting their feelings to the point of uselessness. Without being brutal for its own sake. Just... honest. Accurate. Useful.
That's what the SensAI Triad is built to be.
StorySensAI: The deep-dive diagnostic for your story.
StorySensAI (The Flagship): This is the full diagnostic. It delivers a comprehensive 20-point analysis of your work, highlighting exactly what’s working and what isn't. It finds the cracks in your architecture so you can fix them—delivering the kind of deep, honest critique usually reserved for expensive developmental editors.
SummarySensAI: Context for the AI, promotion for the writer.
SummarySensAI (The Bridge): This provides the "gist" to StorySensAI so it knows the soul of your story while it analyzes the technical details. But it also works for you: it generates a captivating, 75-100 word promotional summary designed to intrigue readers without giving away your plot twists or the resolution.
ConceptSensAI: Testing the foundation of your idea.
ConceptSensAI (The Foundation): This looks at your story idea before you write a single word. Is the concept viable? Is there a real narrative arc here? It helps you find the potential weaknesses in your premise before you spend months writing into a dead end.
And the context flows through all three. SummarySensAI feeds into StorySensAI so that the story review isn't just looking at words in isolation, it understands what the story was trying to be. That distinction matters. A lot. You can read the full story of how StorySensAI got built here.
I've been using it on my own stories for Double Dare!, the follow-up to my first horror collection. Every story goes through the Triad. Every single one. Because I'd rather get a 2.75 in the tool than a one-star on Amazon from someone who actually read it.
What Feedback From People Who Love You Actually Looks Like
You've probably had feedback that left you wondering whether the person even understood what you were trying to do. You've probably had feedback so vague it was useless, or so brutal it just felt like an attack. You've probably sent something to a friend and watched them squirm trying to be kind while being honest, and the kindness won, and you walked away none the wiser.
That's not your fault. That's just what feedback from people who care about you looks like. They're not built to give you the truth about your craft without filtering it through how much they want to protect you.
And it's not just human feedback that fails writers. The AI detection paranoia doing the rounds right now doesn't help either — writers second-guessing every tool they touch instead of focusing on the work.
SensAI doesn't care about protecting you. It cares about the story.
That's the difference.
If you want something that tells you you're brilliant, ChatGPT's right there.
If you want something that finds the cracks in your craft so you can fix them yourself... the Triad's live.
Comments
No comments yet. Be the first!