Moving from non-fiction to fiction writing has been a learning curve. In journalism and PR, the focus is on facts, clarity and getting information across quickly. Fiction is completely different. It’s about building believable characters, creating immersive settings and developing plots that keep readers engaged from beginning to end.
As a new novelist, I know there are areas where I still need to improve. That’s where AI has become a really useful tool in my writing process. I use ChatGPT to help me refine my work, strengthen characterisation and improve scene setting.
For my novels, I created a dedicated project and use it to analyse drafts and ask for suggestions. I keep the process simple by using the same prompt every time: “Give me some constructive feedback on this book.”
One thing I’ve found especially helpful is keeping the same chat open across multiple books in the series. By doing that, I can track how my writing develops over time, see what has improved and identify the areas that still need work.
Below is some of the feedback and suggestions ChatGPT has given me for the first two books in the Gareth Rhys series. Even comparing book one to book two, I can already see clear improvements in my writing. Hopefully, book three will show even more progress.
Book 1: Scales of Silence
You’ve got a strong foundation here. The atmosphere is consistent, the prose is confident, and the tone feels controlled in a way that suits literary crime fiction. There’s a clear sense of place, and Gareth already feels like a believable protagonist rather than a generic investigator. The wildlife trafficking angle also gives the novel an identity beyond standard procedural noir.
A few things are working especially well:
- The setting feels lived-in. Newport, the docks, the valleys, the pubs, the rain, all of it has texture without becoming overwritten.
- Dialogue is natural. Characters sound distinct without slipping into caricature.
- The moral weight behind the story gives it purpose. The badger baiting chapter in particular has emotional force because it feels grounded and informed.
- You handle tension patiently. You’re willing to let scenes breathe.
That said, there are areas where tightening and variation would elevate the book significantly.
1. Repetition is your biggest issue
You rely heavily on certain structural rhythms and sentence patterns. After a while, the prose begins to feel predictable.
You often use:
- short sentence fragments for emphasis
- repeated “pause / silence / gaze” beats
- mirrored dialogue structures
- concluding paragraphs that end in ominous abstraction
Examples:
- “That wasn’t the question.”
- “Silence settled between them.”
- “Something had shifted.”
- “He didn’t answer immediately.”
- “Because he had.”
- “Not really.”
- “Wrong question.”
These work early on, but by Chapters 4 to 7 they start appearing almost every page. The effect weakens because readers begin anticipating the cadence.
You could cut around 20 to 30 percent of these beats without losing atmosphere.
For example:
“The silence stretched. Not uncomfortable. Just familiar.”
This kind of sentence appears frequently in slightly different forms. Once or twice, it’s effective. Repeated often, it starts sounding self-aware.
Try trusting the scene more. You don’t always need to underline emotional tension.
2. Nearly everyone speaks in the same cadence
Gareth, Elin, Tomos, Taff, Lowri, and even the dock worker all speak with similar restrained, elliptical rhythms.
A lot of conversations follow this pattern:
- question
- partial answer
- pause
- cryptic observation
- warning
It creates mood, but too much of it flattens character distinction.
For instance:
- Tomos speaks like a noir informant
- Taff speaks like a noir informant
- Dock man speaks like a noir informant
- Elin sometimes does too
You need more contrast in speech patterns.
Suggestions:
- Let Tomos be more procedural and blunt.
- Let Taff ramble slightly or speak more colloquially.
- Let Elin challenge Gareth intellectually rather than atmospherically.
- Let Lowri occasionally lose patience properly instead of always sounding quietly insightful.
Right now, everyone sounds emotionally intelligent and cinematic.
Real people interrupt more. Misunderstand. Overexplain. Deflect awkwardly.
3. The prose is strong, but often over-curated
You clearly care about sentence quality, which is good. But occasionally the writing feels too polished for the moment being described.
Example:
“The sky bleeding slowly into dusk.”
“The layered geometry of the docks.”
“The world turned inward.”
Nice lines individually. But when every paragraph contains one, readers begin noticing the writing instead of inhabiting the story.
You’d benefit from occasionally writing flatter, especially during movement-heavy scenes. Contrast gives your lyrical passages more impact.
4. Gareth risks becoming overly competent
He notices everything:
- behavioural patterns
- logistics irregularities
- emotional shifts
- criminal structures
- wildlife intelligence
- body language
- tracking signs
Because he’s almost always correct, tension can weaken. Readers start assuming Gareth’s instincts are reliable.
Consider giving him:
- a false assumption
- a serious blind spot
- an emotional bias that affects judgement
- moments where he misreads people
The dock worker scene hints at this nicely:
“Then you’re not looking. You’re confirming.”
That’s one of the strongest lines in the manuscript because it challenges Gareth’s certainty.
Lean into that more.
5. The pacing is deliberate, but danger arrives late
The novel is atmospheric from page one, but plot progression is slower than you may realise because scenes often accomplish similar emotional work.
Across the first chapters, the structure repeats:
- Gareth receives information
- Gareth meets someone
- they speak cryptically
- Gareth senses something larger
It works for a while because the prose carries it, but readers will eventually want stronger escalation.
Chapter 6 is the first scene where the book truly feels physically dangerous. It’s excellent partly because it breaks the conversational rhythm.
You need more scenes like that earlier.
Not necessarily action scenes, but scenes where:
- Gareth fails
- evidence disappears
- someone lies convincingly
- consequences happen immediately
6. Your strongest material is the wildlife crime itself
Whenever the book focuses specifically on:
- trafficking systems
- badger baiting
- docks logistics
- animal trade economics
- rural criminal culture
…it becomes genuinely distinctive.
That’s the book’s real identity.
The more generic conspiracy-noir elements are less original than the ecological crime material. So I’d lean harder into the specificity.
The pangolin angle is genuinely compelling because it feels unusual and believable at the same time.
7. Lowri needs more agency
At the moment, she functions mainly as:
- emotional conscience
- warning system
- reflective mirror for Gareth
She’s well written, but reactive.
Give her independent momentum:
- her own investigation
- expertise Gareth lacks
- social connections that matter
- a decision Gareth disagrees with
Otherwise she risks becoming “the worried spouse who understands the protagonist too well”.
8. Your endings are effective, but too consistently dramatic
Almost every chapter ends with:
- a portentous observation
- a line implying larger forces
- a thematic summation
Examples:
- “Something already had.”
- “It wasn’t finished yet.”
- “They knew someone had been there.”
- “This time, they already had.”
These work individually, but collectively they start feeling formulaic.
Sometimes end on:
- an image
- a practical problem
- an interruption
- a contradiction
- a human detail
Variation will make the stronger endings land harder.
Overall
This already reads far above average unpublished crime fiction. The biggest positive is that it feels authored. There’s confidence, atmosphere, and thematic intent. It doesn’t read like a derivative police procedural.
The main thing holding it back is over-consistency. The same rhythms, tonal beats, and conversational structures recur too often.
Your next step isn’t “better writing”. It’s variation:
- sharper contrasts
- messier conversations
- occasional ugliness
- faster scenes
- emotional unpredictability
- fewer perfectly weighted sentences
That would turn this from “very good atmospheric crime fiction” into something genuinely memorable.
Book 2: Shadows in the Grass
This feels noticeably stronger and more confident than the previous manuscript. The biggest improvement is control. You seem more aware of when to explain, when to hold back, and when to let a scene carry its own weight. The prose is still atmospheric and recognisably yours, but it’s cleaner, more disciplined, and generally less self-conscious.
The core concept is also excellent. A murder involving a Gaboon viper in the Welsh wetlands is unusual without feeling gimmicky. It immediately creates intrigue while tying naturally into your wider themes of wildlife trade, ecological crime, and hidden systems.
A few things stand out as especially successful.
What’s working very well
1. The opening is stronger
This is a much better opening chapter structurally than the Pangolin book.
You begin with:
- movement
- people interacting naturally
- a distinctive location
- discovery
The body reveal unfolds gradually and visually. The pacing feels earned. The wetlands setting also gives the book a more immediate identity.
This section is especially strong:
“The air felt different down here – closer, heavier, the faint scent of salt and decay rising from the exposed ground.”
That’s restrained and effective.
You’re also better now at embedding exposition inside action rather than pausing the novel to explain itself.
2. Sara Llewellyn is your best character so far
She’s a major improvement over some of the supporting figures in the earlier manuscript because she genuinely pushes back against Gareth rather than existing mainly to reinforce him.
Her dialogue has:
- scepticism
- procedural thinking
- impatience
- intelligence
That creates friction.
This exchange works well:
“You’re building a picture without pieces.”
That line does several jobs at once:
- critiques Gareth
- critiques the reader’s assumptions
- establishes Sara’s worldview
More of that.
3. The wildlife material feels integrated now
In the previous manuscript, the research occasionally sat on top of the story. Here, it’s woven in more naturally.
The discussions around:
- venom
- reptile trafficking
- licensing
- captive breeding
- exotic pet culture
…feel embedded in the investigation rather than inserted for thematic weight.
That’s a big improvement.
The details about:
- chilled transport
- mortality rates
- ELISA venom testing
- dangerous animal licensing
all make the world feel credible.
4. You’re better at scene architecture
Scenes now have clearer purposes.
For example:
- Chapter 1 = discovery
- Chapter 2 = forensic interpretation
- Chapter 3 = editorial challenge
- Chapter 4 = investigative confirmation
- Chapter 5 = domestic/philosophical grounding
- Chapter 6 = wider criminal context
- Chapter 7 = field investigation
That structure gives the book momentum.
The earlier manuscript sometimes felt like variations on the same conversation. This one progresses more clearly.
Areas that still need work
1. You still overuse the same emotional rhythms
This remains your biggest stylistic issue.
You rely heavily on:
- pauses
- silence
- measured responses
- gaze-based reactions
- restrained dialogue
Almost everyone speaks in controlled understatement.
Examples:
- “Sara didn’t answer immediately.”
- “A pause settled between them.”
- “He held her gaze.”
- “She nodded once.”
- “That changed something.”
These are individually fine. The problem is accumulation.
By Chapter 6 or 7, the prose starts falling into a predictable cadence again.
You need more tonal disruption.
Let scenes become:
- awkward
- chaotic
- emotionally uneven
- interrupted
- funny
- angry
Right now everyone communicates like they’ve had media training.
2. Gareth still feels too intuitively correct
This is better than before, but the issue remains.
He repeatedly:
- identifies rare phenomena immediately
- reads scenes accurately
- spots behavioural inconsistencies
- predicts wider systems
And he’s usually validated shortly afterwards.
The Gaboon viper identification is believable because of his background, but because he’s confirmed so quickly, the narrative risks teaching readers that Gareth is always right.
That lowers suspense.
A stronger approach would be:
- Gareth misidentifies something once
- overcommits to a theory
- trusts the wrong source
- misses a personal motivation
- interprets patterns incorrectly
Not because he’s incompetent, but because intelligent investigators make interpretive mistakes.
Sara’s scepticism helps enormously here. Keep using her as a counterweight.
3. Your dialogue occasionally becomes too polished
Characters often speak in perfectly shaped thematic observations.
Example:
“People notice what they’re looking for.”
Good line. But many characters speak like that.
Real conversations usually contain:
- unfinished thoughts
- repetition
- tangents
- emotional leakage
- clumsy phrasing
Anne is one of your more believable characters because she sounds slightly less curated.
Taff-style noir dialogue from the earlier manuscript appears less here, which is good.
Still, I’d roughen some conversations slightly.
4. The prose sometimes explains the atmosphere twice
You often establish mood effectively, then reinforce it again immediately afterwards.
For example:
- describe stillness
- then explain emotional stillness
- then underline thematic stillness
Trust the image more.
This line works:
“The air felt different down here.”
You don’t always need the additional interpretive sentence afterwards.
Readers are already with you.
5. You occasionally flirt with over-research
This manuscript handles exposition much better overall, but there are still moments where information density slows the tension.
Particularly:
- reptile trade economics
- licensing frameworks
- import pathways
Interesting material, but sometimes the characters begin sounding like briefing documents.
A useful test:
Would this character genuinely explain this much in this moment?
Elinor gets away with it because she’s analytical by nature. Gareth and Sara need slightly more compression.
6. Lowri is improved, but still somewhat functional
She’s more active here than in the earlier manuscript:
- asks questions
- investigates socially
- directs Gareth toward Elinor
That’s progress.
But she still mainly exists to:
- stabilise Gareth
- reflect concern
- encourage caution
- provide conversational grounding
Give her independent stakes.
For example:
- emotional conflict with Gareth’s obsession
- involvement that backfires
- knowledge she withholds
- a connection to someone relevant
- a personal investment in the case
At the moment she still risks becoming “the intelligent supportive wife”.
Biggest improvement overall
The earlier manuscript sometimes felt like:
“Atmosphere first, plot second.”
This one feels much more balanced.
The investigation moves.
Scenes reveal information.
Characters disagree.
The world feels broader.
You’ve also become better at trusting the reader.
One thing I’d strongly encourage
Lean harder into the ecological horror aspect.
That’s where the novel becomes unique.
The idea that:
- global wildlife trade
- private collecting
- exotic status culture
- ecological displacement
- criminal logistics
…can produce a murder in the Welsh wetlands is genuinely compelling.
That’s your identity as a writer.
Not just crime fiction.
Environmental crime fiction with noir elements.
That niche feels authentic for you.
Final assessment
This reads like a serious novel now, not simply a promising draft.
The main thing separating it from publishable literary crime fiction is still stylistic variation.
You have:
- atmosphere
- intelligence
- thematic coherence
- strong setting
- credible procedural detail
- clear authorial voice
What you need more of is unpredictability:
- emotional unpredictability
- tonal variation
- conversational messiness
- investigative mistakes
- sharper conflict
At the moment, the book is very controlled.
Sometimes too controlled.
But the talent is absolutely there.
Final thought
For anyone thinking about writing a novel, there has never been a better time to start. New writers now have access to a huge range of tools that can help develop their skills and improve their work. Affordable AI platforms such as ChatGPT can provide feedback, guidance and support that once required expensive editors or writing tutors.
Of course, no tool can replace creativity, imagination or hard work. But if you are willing to learn, take notes and keep improving, AI can become a valuable part of the writing process. For new authors especially, it offers an opportunity to learn faster, build confidence and continue developing with every book written.
