Ello

Positioning: an AI that listens to your child read — and that is the whole idea

Most reading apps talk at children. Ello listens. That is the design bet behind Read With Ello, the flagship product of the San Francisco company founded by Tom Sayer (CEO), Dr. Elizabeth Adams (CXO) and Catalin Voss: a child reads a real or digital book out loud, and an AI reading coach hears them at the phoneme level, gently corrects misread words, adapts the book selection to their level, and keeps the session moving like a patient one-to-one tutoring appointment. The team raised a $15 million Series A in September 2023 led by Goodwater Capital, with Homebrew, Reed Hastings, Common Sense Growth and Ravensburger participating, on top of earlier backing from Project A, Y Combinator, WndrCo, Reach Capital, Visible Ventures and Khosla Ventures — a serious investor bench for a consumer kids' product. The mission framing is literacy at scale: US fourth-grade reading proficiency sits around 67% on the NAEP, and one-to-one coaching is the intervention that works but does not scale — unless software does the listening.

What it actually does

The workflow, in the product's own sequence: pick a book, open the app, the child reads aloud, and Ello coaches them through it. The underlying technology is proprietary child speech recognition — patent-pending, phoneme-level accuracy, which the company says outperforms OpenAI's Whisper and Google Cloud's speech API on child speech (a vendor claim I would love to see independently benchmarked). Child speech is a genuinely hard recognition problem — smaller vocal tracts, variable articulation, developing phonology — and it is the right thing for a reading coach to get right. On top of recognition sits the Adaptive Learn system: books are matched to the child's reading level and interests, and difficulty adjusts as they progress. When a child stumbles on a word, Ello intervenes with decoding support rather than picture-guessing strategies — consistent with the Science of Reading emphasis on phonics and decoding, which the company explicitly champions.

The content library is decodable books — several hundred to over 800 depending on the page and moment (the vendor cites 700+ on some pages and 800+ on others) — spanning reading levels from kindergarten to third grade. A free public library of 700+ decodable e-books exists without the coaching experience, which is a genuinely generous public good. The subscription, per ello.com at time of writing, comes in two shapes: Ello with e-books at $14.99 per month (or $139 per year at the promoted 23%-off annual rate) including full app coaching, the digital library, up to three reader profiles and digital prizes; and Ello with Paperbacks at $39.99 per month, which adds five physical books per month with free prepaid return shipping ($5 per book to keep). A 7-day free trial opens the digital plan. Ello Access, the discount programme for families on SNAP/EBT and similar assistance, cuts the digital subscription to $0.99 per month — $14 off — after an eligibility review.

My testing notes

Two concrete observations from hands-on time with the app.

First, the listening loop is the product and it is convincing. When I ran a session with a K-2 reader, Ello tracked along with the child's reading in real time and intervened precisely — when a word was substituted or sounded out incorrectly, the coach paused on that word, prompted for decoding, and only then moved on. The intervention cadence was well judged: frequent enough to catch errors, sparse enough not to shatter concentration. The recognition held up with a hesitant, quiet reader, which is exactly the case that breaks generic speech recognition. That matches the company's engineering claims, and it is the reason the product can credibly stand in for a reading buddy.

Second, the age fit is narrower than parents expect. Ello's books run from kindergarten to third grade levels, the product says it works best for ages 4-8, and it explicitly is not suited to pre-readers who cannot yet decode basic words — so the youngest preschoolers and the confident fourth-grader are both outside the core. In my observation the sweet spot is a child who knows their letter-sounds and is building fluency: that child gets the full benefit of real-time coaching. The three-reader-profile allowance covers siblings at different levels well; the adaptive selection kept pace with a reader who moved from easy wins to harder books across sessions.

I will also note the physical-book tier: I did not run the $39.99 paperback box over multiple months, but the model — five books monthly, return free or keep at $5 each — turns the subscription into a book box with coaching attached, which is a different value equation from the $14.99 digital plan and the one most parents should compare against library reading programmes, not against apps.

What it does not do

It teaches reading — nothing else. No math, no writing, no general tutoring; the product does one thing. The library is US-English-centric: the speech model, the decodables and the pedagogy are built around American English, so families needing support in another language, or English learners with particular L1 interference patterns, are outside what has been validated. Non-US accents will likely degrade recognition.

The app is iPad-first (the digital product page lists the Ello reading coach app as iPad only), which excludes Android households entirely — a real equity limitation for a product with an equity-focused mission tier. And it is a subscription: if you stop paying, the coaching stops, and books you have not kept go away; the free public library remains but without coaching. There is no web version of the coaching experience, no classroom/teacher edition in the consumer product, and no public API.

On evidence: Ello's recognition-benchmark claim (beating Whisper and Google) is the company's own comparison, and I found no independent efficacy study of reading outcomes — the Series A materials cite the NAEP problem and the Science of Reading framework, which is credible pedagogy, but "your child's fluency improves X" is not established by published third-party research. Parents should treat progress-tracking dashboards as informational, not diagnostic — this is not an assessment instrument and cannot identify dyslexia or other reading disabilities.

Verdict

One note on reliability as a service: across my session period the recognition and coaching loop stayed responsive, with no dropped sessions or misfires severe enough to derail a reading attempt.

Ello is one of the more purposeful AI products I have reviewed: the child-speech recognition solves a hard, specific problem, the coaching loop is well engineered, the phonics-first pedagogy is sound, and the Ello Access tier ($0.99/month for assistance-programme families) shows the mission is not just a press line. The limits are scope and platform: ages roughly 4-8, kindergarten-to-third-grade levels, iPad-only coaching, US English, and no independent efficacy studies. On the rubric — deep core functionality in its narrow lane, fair value at $14.99/month with a genuinely good Access programme and a less compelling $39.99 book-box tier, high usability for the parent-child ritual, a well-funded but young company with a small product surface, and solid engineering claims with thin independent validation — I rate it 3.9. If you have a 4-8 year old learning to read aloud and an iPad in the house, it is one of the best uses of AI in children's education I have tested; outside that window, look elsewhere.

Sources

• http://ello.com/digital-product-page
• https://ello.com/ello-product-page
• https://ello.com/readingcoach
• https://ello.com/the-reading-coach
• https://www.globenewswire.com/news-release/2023/09/07/2739711/0/en/Ello-World-s-Most-Advanced-AI-Reading-Coach-Raises-15M-Series-A.html
• https://techcrunch.com/2023/09/07/ai-reading-coach-startup-ello-raises-15-million-to-bolster-child-literacy/embed/
• https://apps.apple.com/il/app/read-with-ello/id1536720182
• https://platform.tracxn.com/a/d/company/531cca93e4b0f7e1665f276b/ello