Which AI tools cover the whole self-publishing path?

The short answer

Self-publishing has four stages — writing, cover, formatting, storefront — and the best setup assigns one proven tool per stage: EarnDraft for the manuscript, Canva's free tier for the cover, Kindle Create or your pipeline's exports for formatting, and KDP plus a direct storefront for sales.

The only stage where AI does the heavy lifting is writing. Covers need your taste, formatting needs checklists, and storefronts need decisions about exclusivity and pricing that no software can make for you.

Reviewed below: the best pick per stage, the closest alternative, where each stops, and what each costs — with sources checked in September 2026.

1. Writing: EarnDraft (manuscript, cover draft, files)

The writing stage is the bottleneck that kills most books, so it gets the strongest tool: describe your idea and receive a structured draft with a designed cover and PDF plus DOCX exports, free to preview with no credit card. Pro unlocks the exports and the live page.

Chatbot assembly can't compete at this stage for one reason: the manuscript must arrive as chapters, not chat logs.

Whatever drafts your words, budget the review pass: facts checked against primary sources, your stories inserted, voice fixed aloud. Our verification workflow and editing pass cover the two hours that separate a draft from a book.

Not for this stage: marketing copy tools and paraphrasers. They polish sentences; they do not produce manuscripts. Buy them later for launch copy if at all.

2. Cover: Canva (best free cover designer)

Covers sell books at thumbnail size, and Canva's free tier remains the best zero-cost way to make one: templates sized for ebooks, typography that behaves, and exports large enough for KDP's uploader. Amazon wants a high-resolution JPEG with readable title text at thumbnail scale — design small, not large.

AI image generators can supply backgrounds and textures, but keep a human in charge of the title treatment: legibility at 100 pixels wide is a design decision, not a generation prompt. If the title is unreadable small, the cover fails whatever it looks like large.

When to pay: if the book carries your business reputation, a freelance cover is the highest-leverage money in self-publishing — collect quotes, check portfolios at thumbnail size, and brief the designer on the promise, not the plot. Conceded plainly: no pipeline-generated cover beats a good human designer; ours gets you published, theirs gets you clicked.

3. Formatting: Kindle Create plus your pipeline exports (free)

Amazon's own Kindle Create tool is free and converts a clean manuscript into a Kindle-ready file with a clickable table of contents — it appears in KDP's own tools and resources pages, which is the strongest endorsement a formatting tool can get.

The input matters more than the tool: one font, real heading styles, chapter breaks as real breaks, images simple and centered. Pipeline exports from EarnDraft arrive pre-structured, which is why the writing-stage choice echoes into this stage.

KDP accepts DOCX, EPUB, and PDF manuscript uploads and converts on upload; DOCX from a clean source converts most reliably, while PDFs preserve fixed layouts but reflow poorly on small screens. Then preview on KDP's own previewer across phone, tablet, and e-ink before submitting — our Kindle formatting checklist walks the whole pass.

4. Storefronts: KDP for discovery, direct for margin

Uploading to KDP costs nothing; Amazon takes its share per sale — 35% or 70% royalty options depending on price band and territory, with the 70% band at $2.99–$12.99 on Amazon.com at last check. A $4.99 ebook at 70% earns roughly $3.45 per sale after the small delivery deduction.

Pair it with a direct storefront: Gumroad charges 10% plus 50 cents per direct sale with no monthly fee (30% for sales through its Discover marketplace), so the same $4.99 book keeps about $3.99 direct. Etsy brings search traffic at $0.20 per listing plus 6.5% and payment processing — roughly $4.07 kept on $4.99 by our math.

The exclusivity question decides the mix: KDP Select's Kindle Unlimited enrollment demands 90-day digital exclusivity to Amazon in exchange for page-read income and promo tools. If the book feeds a course or newsletter, stay wide; if you have no audience, test one exclusive term. Our Gumroad-versus-KDP guide runs this decision with full math.

Side by side: the full path and its costs

Four stages, four best picks, one honest total: the DIY path runs from free to one modest subscription month, while outsourcing cover and editing moves the total into the hundreds — with quotes, not guesses, since freelance rates vary too widely to print.

The stage most authors underfund is review: the verification and editing passes cost time rather than money, and skipping them is how factual errors and robotic prose reach paying readers.

Methodology note: writing tools were compared hands-on against the same four questions as our generator ranking; storefront and KDP figures were checked against official fee and help pages in September 2026 and are hedged where vendors revise terms. Confirm before you price.

• Cheapest finished book: free tiers plus KDP uploading — under one subscription month.
• Highest-leverage spend: human cover, then human proofread.
• Never skip: disclosure answers, fact verification, previewer check.

Which stack should you assemble?

First book, no audience, minimal budget: EarnDraft free start, Canva cover, Kindle Create formatting, KDP wide plus Gumroad direct. Total cash outlay before your first sale can genuinely be zero — the honest-scope caveat being your evenings, which are real money too.

Business book feeding an offer: same stack plus a freelance proofread and a stronger cover budget, sold wide so the funnel keeps its direct-sales page. The book is a customer-acquisition cost; account for it like one.

Whatever the stack, publish in this order: finish the manuscript, verify the facts, lock the cover at thumbnail size, preview every file, then answer disclosure honestly. The weekend plan compresses exactly this sequence into two days for short books.

The forgotten stage: editing and review

Between formatting and storefronts sits the stage most stacks omit: review. AI drafts arrive fluent and unchecked, so plan three passes — verification of claims against primary sources, a voice pass adding your stories and rhythm, and a final proofread on the exported PDF, where formatting reveals awkwardness the editor hid.

Do the first two passes yourself; they are where the book becomes yours. The claims-log method from our accuracy workflow takes about two evenings for a short book, and the humanizing edit about two hours. Neither is delegable to the tool that produced the draft — self-review of AI output is like proofreading your own typing, necessary and insufficient.

Hire the third pass when the book carries weight: a freelance proofreader for business books, a qualified specialist for health, legal, or financial content. Get per-project quotes, brief on claims rather than prose, and treat the invoice as insurance rather than expense — one caught error repays it.

Read the finished PDF start to finish on a phone before submitting anywhere. Most buyers meet your book on mobile; a proofread that only happened in a desktop editor missed the actual reading experience.

Launch sequence: the first 30 days

Week one is verification, not promotion: buy or borrow your own copy on each storefront, confirm the real files match the previews, fix anything the previewer hid, and point your live book page and profiles at the listings. Early errors fixed fast never become one-star reviews.

Weeks two and three are deliberate first readers: your list, your communities, launch pricing, and honest review requests to people who actually read the book. Storefront algorithms notice velocity and conversion — ten real readers now beat a hundred promised someday-readers.

Week four is the review date you set before launch: copies per shelf, margin per shelf, traffic sources, and reader questions. Reader questions are the product roadmap — they dictate the second edition, the sequel, and the lead magnet that feeds the funnel. Our side-hustle expectations sets the numbers to expect while this compounds.

Throughout, keep one channel log: where each buyer came from, in one place. Attribution done casually for thirty days beats analytics installed perfectly and never read — and it decides your next book's launch plan.

After the ebook: print and going wide

Once the ebook sells steadily, print becomes the expansion: paperback editions through KDP's print path reuse the manuscript with print formatting (margins, bleed, and trim sizes the ebook never needed), and buyers who prefer paper pay premium prices for the same words. Start the print edition with earnings, not savings.

Going wide means carrying the ebook beyond Amazon: direct sales stay (Gumroad keeps margin and emails), Etsy adds discovery for planners and workbooks, and wideretailers enter only if your readers shop there. Each new shelf needs its own cover-size check and description formatting — small work, scheduled rather than squeezed.

The ordering principle: prove demand on one shelf before multiplying shelves. Every additional storefront multiplies listing maintenance, review monitoring, and price-sync chores — worthwhile for a proven book, pure overhead for an unproven one.

Honest-scope note: print formatting, translations, and audiobooks are separate crafts this guide doesn't cover. The path above gets the ebook earning; expansions ride on its evidence, in the order readers request them.

Categories and keywords that place the book

Categories are shelf placement, not self-expression: browse the Kindle store as a buyer, find where comparable books rank in the top hundred, and choose those categories over grander-sounding ones where giants dominate. A focused book ranking visibly in a small category outsells an invisible one in a large category every time.

Keywords follow buyer language, not author language: the phrases readers type when they feel the problem, harvested from autocomplete, community questions, and competing titles' reviews. Seven keyword slots filled with reader phrasing beat seven slots of clever synonyms — match the search, then earn the click with the cover.

Revisit both quarterly: categories drift as Amazon reshuffles, competitors move, and your reviews accumulate into new positioning. Fifteen minutes with a buyer's eyes each quarter keeps the book on the shelf where its readers actually browse.

For series or multi-book catalogs, coordinate rather than duplicate: adjacent categories per book, shared keyword roots, and back matter cross-linking so each title's traffic feeds the others. One book's rank becomes every book's discovery.

Honest-scope note: nobody outside Amazon knows the full ranking algorithm, and anyone selling certainty about it is guessing for money. Placement plus cover plus reviews is the known game — play it well and skip the gurus.

The cover checklist that earns the click

Design for the thumbnail, judge at 100 pixels wide: title legible, one focal element, contrast that survives compression. Full-size beauty that dissolves into gray mush at listing size is the most common cover failure, and the preview that catches it takes ten seconds.

Promise, don't decorate: the title names the outcome, the subtitle names the reader or the mechanism, and every visual element serves recognition at speed. Compare against the top ten covers in your categories — distinct enough to spot, familiar enough to trust.

Test before committing: show three variants to ten people matching your reader sentence, ask which they'd click and why, and listen for the why more than the vote. One round of testing beats three rounds of designer revisions aimed at taste nobody measured.

Keep the source files and a variant folder: seasonal tweaks, second-edition badges, and A/B tests all need editable originals. A cover you cannot revise is a cover frozen at your least-informed moment.

• Legible at 100px, one focal element, real contrast.
• Title promises, subtitle qualifies, visuals serve speed.
• Test with ten real readers before finalizing.

Validate before you publish, not after

Publishing unvalidated is the most expensive mistake in self-publishing: finished files, listed products, and zero demand discovered at the worst possible moment. The four demand checks cost one weekend — stranger searches, adjacent spending, unprompted questions, and a landing page pulling emails — and they belong before the writing stage, not after the storefront stage.

Read competing reviews as product specifications: three-star reviews on adjacent books list exactly what buyers felt was missing, which becomes your differentiation chapter by chapter. A book engineered against documented complaints launches with positioning no marketing budget can buy.

Pre-sell the promise while drafting: share the title and one-chapter preview with the communities where you validated, collect emails on the live book page, and let signup velocity set the launch print run of your effort. Fifty signups from the right readers outranks five hundred followers who never asked.

Kill or narrow fast when signals stay dark: no searches, no adjacent spending, no questions, no signups. A shelved idea costs a weekend; a published book nobody wants costs months of hope plus the motivation for book two. Validation protects the catalog, not just the title.

The pipeline makes validation cheap to act on: a passing idea drafts the same week it passes, while the evidence is fresh and the reader language still rings in your ears. Speed from signal to draft is the underrated advantage of the whole stack above.

Sources checked (September 2026)

Storefront and KDP figures verified against official pages this month; freelance-rate ranges deliberately omitted — collect quotes.

• KDP AI-content rules: https://kdp.amazon.com/en_US/help/topic/G200672390
• KDP digital pricing (35%/70%): https://kdp.amazon.com/en_US/help/topic/G200634500
• KDP Select terms: https://kdp.amazon.com/en_US/help/topic/G200798990
• Gumroad pricing: https://gumroad.com/pricing
• Etsy fees: https://www.etsy.com/legal/fees/