Veritas reads your textbook, notes or any document to you in your own voice, every word highlighted. Then it checks what you understood and brings back what you missed, until you’ve mastered it.
Public release · Q1 2027
to the start of Q1 2027
Launch gates measured on a review set of 400+ real chapters across 12 subjects.
// The story · 2:32
Watch it all with sound, or jump to a chapter.
[ 01 ] // The problem
Unless you learn it differently. What we read fades fast when we never practice it. More effort isn’t the answer. A better method is.
Bone et al., iScience, 2025 · Roediger & Karpicke, Psychological Science, 2006 · Dunlosky et al., 2013
[ 01b ] // The problem in detail
A 2025 study in iScience found that the share of Americans who read for pleasure on a given day fell from 28% in 2004 to 16% in 2023. A 2026 Cornell report says students raised on digital technology are quick to find information but struggle with sustained reading and complex arguments.
In a 2013 review of ten common study techniques, Dunlosky and colleagues rated rereading, highlighting and summarizing as low utility. Practice testing and spaced practice rated high.
Ebbinghaus described the forgetting curve in 1885, and Murre and Dros replicated it in 2015. Material you never revisit fades quickly, and each well-timed review flattens the curve.
Put together, the evidence points away from faster reading and toward practice. With the same passage and the same study time, students who read it once and recalled it three times remembered 61% a week later. Students who reread it four times remembered 40%.
The methods that work are testing yourself, spacing your practice and explaining things in your own words, and they are hard to keep up alone. Veritas handles the dull parts. It writes the questions from what you read, finds the weak points, and brings them back a little before you would have forgotten them.
The tools people use today stop at the first step. Reading apps stop at listening, and study apps live somewhere else and never see what you read.
Bone et al., iScience, 2025 · Cornell University, Report of the Committee on the Future of the American University, 2026 · Dunlosky et al., 2013 · Ebbinghaus, 1885 · Murre & Dros, PLOS ONE, 2015 · Roediger & Karpicke, 2006
[ 01c ] // Who it is for
Three groups read for different reasons and put up with different amounts of effort. Veritas is built for the first one first, because the need is urgent and a habit forms most easily when there is a date.
A fixed deadline, a lot of text and not enough time. Listen to the chapter on the bus, take the check when you sit down, and review your weak points until the exam. Textbooks, lecture notes and papers.
A report or a technical PDF to understand quickly and still know next quarter. Hear it at the fastest speed you still understand, write what matters in your own notes, and fit review around work.
The books you always meant to finish. Work through them and come away able to explain each one to someone else.
[ 02 ] // The product
Long-term memory has a recipe. Veritas runs all three steps on anything you read, in one place.
The library takes PDFs, EPUBs, Word files, web links, photos, scans and pasted text, organized on shelves and in collections. Chapters are found for you.
Read six short passages aloud once, about two minutes. After that, every book can be read to you in your own voice, with every word highlighted. Prefer a preset voice? That’s there too.
People overestimate how fast they can listen and understand. A short test reads passages at rising speeds, each followed by a comprehension question, shows where understanding starts to fall, and how long the chapter and the book will take at that pace.
A picture of the idea builds beside the text, following Mayer’s multimedia principle. Every chapter ends with a knowledge check of about 12 questions that finds your weak points, then you practice those by drawing, teaching it back, flashcards, quizzes or games.
Today lists what’s due, how many questions and about how long, before you start. What you miss comes back sooner. What you know comes back less often. A typical review: 25 questions, about 20 minutes, across three books.
[ 02b ] // Practice
[ 02c ] // More of the product
[ 02d ] // Architecture
[ 02e ] // How it works, step by step
You only ever use the web app in your browser: the library, voice setup, read-along, check and practice, and the daily review. Sign-in is hosted, so Veritas never stores a password. The browser records your voice clips, and it asks for consent before it does.
Everything else runs in the cloud. The web server checks who is signed in, shows each learner only their own data, and hands slow work to a job queue. The queue is a table in the database, so jobs survive a restart. Four workers take jobs from it: import, which handles scans and Word files; voice, for preset voices and your own; timing, which finds the moment each word is spoken for read-along; and the study model, which writes and grades questions.
The learning loop runs on top. A check of about 12 mixed questions closes each chapter, and written answers are graded against the source passage. What you miss feeds six ways to practice, and a spaced scheduler brings it back in one daily review across everything you read.
The daily review ran in the closed beta. The fuller learning engine, with a learner profile, the scheduler and mastery tracking, is still being built out.
[ 02f ] // Where the AI is
Most of Veritas is ordinary software. The AI does four jobs, and what it is allowed to do in each was decided in advance.
| Feature | What the model does | The rule it works under |
|---|---|---|
| Your-voice reading | Builds a voice from your recording | Only your own voice, from a recording you make on purpose. Preset voices are the alternative. |
| Questions and flashcards | Writes questions from your text, including diagram questions | A question must be answerable from that text, and every answer is explained. |
| Feedback on answers | Says what a written, drawn or spoken answer covered and missed | Compared against the source passage, not against general knowledge. |
| Review schedule | Decides when weak points come back | You see what is due and how long it will take before you start. |
The main risk in a product like this is a confident wrong answer: a question with the wrong key, or feedback that calls a correct explanation incorrect. A learner might not notice and would end up studying the mistake. So evaluation is a launch requirement. Before any question type or feedback type ships, it has to pass a review set built from real chapters. The set checks whether the answer is in the text, whether the key is right and whether the explanation helps.
After launch, the measures change from a review set to real use: flags per 1,000 questions, the rate of disputed feedback, mastery gain per book, and how many learners return to the daily review in week four.
[ 03 ] // Decisions
Every format is practice testing, generative learning or spacing. If an idea is popular but doesn’t strengthen recall, it waits.
Summaries are the most requested feature, and the research rates summarizing low for remembering. Each chapter starts with an empty note.
What you miss in a check comes back in practice and in review. That connection is the product, so a feature gets cut before the loop breaks.
[ 04 ] // From six apps to one
None of them knows what you missed. Veritas connects listening, notes, checks, practice and review on the same text.
| What you need | Veritas | Read-aloud apps | Note tools | Flashcard and quiz apps |
|---|---|---|---|---|
| Reads any document aloud, every word highlighted | Yes | Yes | No | No |
| In your own voice | Yes | Some, paid | No | No |
| A knowledge check every chapter | Yes | No | Partly | No |
| What you miss comes back in practice and review | Yes | No | No | No |
| Spaced review across everything you read | Yes | No | No | Cards you made |
| A plan with a finish date for every book | Yes | No | No | No |
[ 04b ] // How it is tested
Veritas is not publicly released. These numbers come from a waitlist, a closed beta and an offline evaluation, not from a public launch.
| What it is | Number | How to read it |
|---|---|---|
| People on the waitlist | 2,500+ | Signed up for early access, with no paid marketing. Interest, not usage. |
| Pilot partners | 3 | University study-skills centers signed on for the first pilot. There are no pilot results yet. |
| Closed beta | 60 | Students who used the product before launch. A small group: an early signal, not a measurement of learning. |
| Returned for daily review in week 2 | 85% | Of the beta students. Coming back to the daily review is the habit the product depends on, and week 2 is early. |
| Voice naturalness | 4.4 / 5 | How the beta students rated their own cloned voice, close to how they rated their own recordings. |
| Source-grounded answers, answer-key accuracy | 97% / 99% | Offline evaluation on 400+ real chapters across 12 subjects. It checks that answers come from the chapter and that the key is right. It doesn’t measure whether learners remember more. |
Still to come before launch: a first pilot with real learners against a rereading baseline.
[ 04d ] // Questions people ask
PDFs, EPUB books, Word documents, web pages, photos and scans, or text you paste in. Everything goes into one library with shelves and collections, and chapters are found for you.
Yes. You read six short passages aloud once, about two minutes, and Veritas builds your voice from them. You can also choose a preset voice, and change speed at any time.
You decide. In your plan you set how many books to master a week or a month and which days you study, and it shows the minutes a day that takes. A chapter’s knowledge check is about 12 questions, about 12 minutes.
What you miss in a chapter check becomes a weak point. Weak points come back in practice and in your daily review, sooner while you are still getting them wrong and less often once you know them.
No. There are six ways to practice: written answers, multiple choice, drawing from memory, teaching it back out loud, flashcards and games. Veritas recommends one from your own results, and you can reorder the list.
Keep them if they work for you. Veritas is for when you want listening, notes, checks and review on the same text, so what you miss is what comes back.
You do. Notes start as a blank bullet for each chapter and stay entirely yours.
Pricing will be announced before launch. People on the waitlist hear first.
The public release target is Q1 2027, open to the waitlist and to the public, starting with students who have an exam ahead.
[ Au ] // The gold standard, here
Every Aurifex product must solve a real problem, be proven before it ships, and be effortless to use. Here is what that means for Veritas.
We forget most of what we read, and the habits that feel productive, like rereading and highlighting, help least.
No question or feedback type ships until it passes 400+ real chapters across 12 subjects: 97% source-grounded, 99% answer-key accuracy.
A short speed test and mode test set the defaults from your own results, and one daily review covers all your books.
[ $ ] // Outlook
| Year | 2027 | 2028 | 2029 | 2030 | 2031 |
|---|---|---|---|---|---|
| Expected net revenue (current trajectory) | $120K | $690K | $1.9M | $3.4M | $4.9M |
| Up to | $150K | $890K | $2.6M | $5.2M | $8.0M |
[ 05 ] // Release
Research, the design of every step, decisions on where AI belongs and the launch plan.
No question or feedback type ships until it passes 400+ real chapters across 12 subjects.
Open to the 2,500+ people on the waitlist and to the public, starting with students who have an exam ahead.
[ AURIFEX ] // More from Aurifex
Veritas is an Aurifex product. Every Aurifex product is held to the same gold standard for AI in skilled work.
info@aurifex.com