Visual Summary / DVFZ Books
Forty tools, nine jobs, and the judgment that stays yours
A 2023 directory of AI writing tools, read for the part that lasts: the jobs, not the products.
Scroll to begin ↓
01 The premise
AI does the technical work, and the judgment stays with you
The book spends one short chapter on what AI is, and it uses that chapter to narrow the claim rather than to widen it. AI here is software that handles the mechanical parts of writing so your attention goes to the parts only you can do.
It describes AI as a second pair of eyes that catches typos and suggests better phrasing, and it says the value goes past spell checking: analysing tone, suggesting structure, offering sources and keywords. It reports a Tyton Partners survey of 2,000 two-year and four-year college students in which “nearly half of students said they have used generative AI writing tools at least once; 14% considered themselves occasional users (i.e., monthly); and 13% were frequent users”.
Then it draws the line that the other nine chapters keep behind them. “Your unique voice, creative flair, and ability to connect with readers are all you.”
Let’s be clear, however, that AI is a tool, not a magic wand. While AI can write your novel, it lacks a human touch. Introduction: The Rise of AI in Writing
Write one sentence naming the writing problem you actually have. The book’s closing instruction is to “start by identifying your specific writing needs and challenges”, because the category of tool you need follows from that sentence, not from the product you happened to hear about.
02 How the book is built
The tools are filed by the job they do, not by the brand
Forty tools, nine categories, one chapter each. The organising move is that every chapter is a job in the writing process, and the tools inside it are competing answers to that same job.
The jobs are content generation, text rewriting, summarization, fact-checking, plagiarism checking, AI detection, writing style improvement, proofreading and grammar checking, and a final chapter of multifunctional tools that cover several jobs at once. That last chapter is by far the largest, with fourteen entries, which tells you where the market was heading when the book was written.
This is why the shape of the book outlives its contents. Products get renamed, repriced and retired. The nine jobs stay exactly where they are.
Write out your own last writing project as a list of steps, then mark which of the nine jobs each step belongs to. Most people find two or three jobs doing all the damage, and those are the only chapters they need.
03 Drafting
Generating text and rewriting text are different jobs
The first two tool chapters look alike and are not. Content generation starts from a prompt or a keyword and produces material that did not exist before. Rewriting starts from text you already have and changes how it reads while holding the meaning steady.
On the generation side the book lists Article Forge, which it says “specializes in generating complete articles based on the keywords or topics you provide”, alongside Bard, Copy AI, Copysmith, INK AI and Writesonic. On the rewriting side it lists QuillBot, which it describes as refining “sentences while preserving their meaning”, along with Wordtune, Word AI and Rewriter.
Knowing which of the two you need saves the most common wasted hour: generating a fresh draft when the draft you had only needed loosening, or paraphrasing a page when you had nothing worth paraphrasing yet.
Before you open anything, answer one question: do I have text on the page or not? Nothing on the page sends you to chapter 2. Text on the page that reads badly sends you to chapter 3.
04 Reading the sources
A summarizer tells you what a source says, and you still have to open it
Chapter 4 covers the smallest category in the book, three tools, and it is the one aimed squarely at people who have to read before they write.
Scholarcy, the book says, “reads research articles, reports and book chapters and breaks them down into bite-sized sections, highlighting key information such as study participants, data analyses, main findings and limitations”, and generates flashcards from them. Wizdom AI searches publications, annotates PDFs and manages citations. scite works on a different axis: its Smart Citations classify each citation as “supporting, contrasting, or just mentioning the cited work”, so you see how a paper has been received rather than only how often.
Notice what the extracted fields are for. Participants, analyses, findings and limitations are the things you check a claim against. The summary is a route into the source, not a replacement for it.
When a summarizer hands you a finding you want to use, open the original and read the limitations it pulled out. If you are citing, check in scite whether the citations to that paper support it or contrast with it before you lean on it.
05 Before you publish
Three checks answer three different questions, and none substitutes for another
Chapters 5, 6 and 7 look like one topic and are three. They ask whether the writing is true, whether the words are yours, and whether a machine wrote the passage.
Fact-checking asks about truth. ClaimBuster, built by a team led by Chengkai Li at the University of Texas at Arlington, “automatically identifies statements or claims within your text that require fact-checking”, cross-references them, attributes the source and scores its credibility. Factiverse, Filtir, Full Fact AI, Adverif.ai and Perplexity AI fill out the chapter.
Plagiarism checking asks about authorship. Unicheck “scans texts against various sources, such as internet databases and academic papers, to identify and highlight similarities”, and Copyleaks does the same across billions of pages. AI detection asks a third thing entirely: Originality AI “detects text produced by generative AI models such as GPT-3 and GPT-4” and returns an originality score and report.
Keep them apart in your head. A clean similarity report tells you nothing about whether a sentence is true, and a passing fact check tells you nothing about where the words came from.
Name the risk in the piece you are working on before you pick a checker. Wrong facts, borrowed words and undisclosed machine drafting are three separate exposures, and each has its own chapter.
06 Editing and voice
Editing tools disagree about what a better sentence is
Run one paragraph through several of the book’s editing tools and you get different edits, because each one is built to improve a different thing.
Hemingway Editor works on readability and structure, “highlighting passive voice and adverbs, and detecting hard-to-read sentences”. WordRake, built in 2012 by the lawyer and legal writing expert Gary Kinder, applies “over 35,000 editing algorithms” to make writing “more concise, clear, and readable”. DeepL Write offers real-time suggestions to improve “clarity, coherence, and impact”. ProWritingAid produces reports on “readability, sentence structure, and overused words” and adapts its suggestions to academic, business or creative writing.
Two of them leave the sentence behind. Textio “focuses on your words’ impact on the target audience” rather than on grammar and style. Acrolinx aligns a whole team’s writing with an organisation’s “brand voice and guidelines”, and Copy AI lets you save custom brand voices to generate against.
None of that decides what you sound like. The tool proposes an edit against its own definition of better, and you are the one who knows whether that definition is the right one for this piece.
Name the kind of writing first, then pick the editor built for it. ProWritingAid changes its suggestions depending on whether you tell it the piece is academic, business or creative, and choosing that setting badly is why good tools give bad advice.
07 Choosing, and dating what you choose
Every tool page is a snapshot, so compare on the columns and then trial it
Each chapter ends with a chart summary, and the columns of that chart are the book’s real selection criteria: type of writing supported, core features, platform compatibility, user experience, pricing, and whether a free trial is available.
The book is unusually direct about how long that information stays good. “The information presented in this book is accurate as of September 2023.” It adds that the authors “cannot control changes made by the creators of each AI tool discussed” and that “readers are advised to verify the data and the AI tools before relying on them in their writing endeavors”.
Its recommended sequence survives that expiry date. Identify the need, shortlist on the six columns, then test rather than read. “Take advantage of the free trials and freemium versions many AI tools offer. Dive in, experiment, and explore without the fear of making mistakes.”
It is crucial to consistently remember that AI tools are designed to assist, not replace, humans. They are not infallible, but they certainly streamline processes and enhance efficiency. Conclusion
Treat every price, plan name, company and feature in the book, and on this page, as a September 2023 reading. Check each one on the tool’s own site before you commit, and expect some of the forty to have changed hands, changed names or gone.
08 The whole book
Pick the job, run the check, then trial the tool
The forty product entries will date. The three moves underneath them will not, and they are what the book is actually teaching.
Nine categories
Generate, rewrite, summarize, fact-check, check originality, detect AI, improve style, proofread, or all of the above. Choose the chapter before you choose the product.
Three questions
Is it true, are the words yours, was it machine-written? Different tools, different reports. Clearing one gate says nothing about the other two.
Trial, then trust
Shortlist on the six chart columns, use the free trial, and treat every price and feature in the book as a September 2023 reading that needs checking.
The book’s own summary of the whole forty is one sentence: AI tools are designed to assist, not replace, humans. Everything on this page follows from taking that literally.
Adapted from 40 Powerful AI Tools for Writing
HumAI Co-Lab Team, edited by Selynna Payne · © 2023 DVFZ Training Consultancy Services
Nearly 60 cited sources in the source edition · tool data current to September 2023