Practice Me: Revenue, team and how it was built
Practice Me (iPhone and web app for spoken English practice with AI tutors, Not stated in the sources): US$4.6K verified revenue in 30 days, October 2026. Evidence grade A: disclosed by a company or a filing, or verified by a payment processor. Team: 1 named founder, Guillaume Sarfati, who sells the app under their own name on the App Store; headcount not published. Launched 3 September 2025; still running, revenue flat.Added 27 October 2026
- Outcome
- US$4.6K verified revenue in 30 days, October 2026Revenue flat
- Evidence
- A disclosed by a company or a filing, or verified by a payment processor
- Launch
- 3 September 2025
- Status
- Still running
- Team
- 1 named founder, Guillaume Sarfati, who sells the app under their own name on the App Store; headcount not published
- Outside funding
- None reported
- Group
- Learning and study
- Main channel
- Search and AI answers
Practice Me is an app for speaking practice: the learner talks out loud in English and gets a spoken answer, instead of reading or writing exercises. The AI tutor listens to the learner, writes an answer, and reads it aloud in a native American or British voice. The App Store page lists pronunciation feedback, interview and business English, IELTS and TOEFL practice, live subtitles with translations into the learner's own language, and a word list built from each conversation. The App Store compares the price with human tutors, which it says "often" cost "$15–30 per 45 minutes". No source names the AI models behind the voice or the answers; an update of 28 March 2026 says only "Upgraded AI model for better accuracy and fluency".
The App Store lists Guillaume Sarfati as the developer and seller; the copyright line names a company, CODELINE TEKNOLOJI, which no other source explains. Version 1.0 was released on 3 September 2025. For seven months the app made almost nothing: TrustMRR, a site that shows startups' revenue read directly from their payment accounts, records US$62.78 from November 2025 to January 2026 and no revenue in February or March. Revenue rose in April 2026, and from May it was between US$3,232 and US$4,677 a month. On 27 October 2026 TrustMRR, reading RevenueCat (a service that records App Store subscription payments for app makers) and Stripe, showed US$4,560 of revenue in the last 30 days, monthly recurring revenue (MRR), the subscription income that repeats every month, of US$3,044 from 235 active subscriptions, and US$26,713 since launch.
The revenue figures are grade A: TrustMRR and a second site, whatsthe.app, both read them from the app's payment accounts, and whatsthe.app showed an MRR of US$2,910 on 1 October 2026. The founder's own fields, such as the marketing channels "App store optimization" and "SEO" and a profit margin of 68%, are grade C. The page also shows four earlier companies that sold English practice with human teachers and failed, and notes where the app's own website shows a selected App Store rating.
How much revenue does Practice Me make?
Each row shows the evidence grade and the source numbers from the list at the end of this page. The grades are explained in How Aiolum verifies.
| Fact | Value | Grade | How we know | Sources |
|---|---|---|---|---|
| Revenue in the last 30 days, to 27 October 2026 | US$4,560, down 7.7% | A | TrustMRR, revenue synced 27 October 2026 from RevenueCat and Stripe through API keys. Calculated: US$54,720 a year at that rate. The last 3 months brought US$13,571 and the last 6 months US$24,813. | [1] |
| Monthly recurring revenue (MRR), 27 October 2026 | US$3,044 from 235 active subscriptions | A | TrustMRR snapshot; MRR growth in the last 30 days 0.0%. Calculated: US$12.95 of MRR per subscription per month, and US$36,528 a year at that rate. | [1] |
| Second tracker, 28 days to 1 October 2026 | US$3,913 of revenue, MRR of US$2,910, 211 active subscriptions, 38 active trials | A | whatsthe.app: "All data is verified through RevenueCat. Last updated: Oct 1, 2026". Its figures are labelled as RevenueCat data, and it does not mention the Stripe web payments. Calculated: TrustMRR's MRR 26 days later is 4.6% higher and its subscription count 11.4% higher. | [2] |
| Revenue by month | US$32.80 in November 2025, US$19.99 in December, US$9.99 in January 2026, none listed for February and March, US$1,829 in April, US$3,630 in May, US$3,232 in June, US$4,529 in July, US$4,677 in August, US$4,461 in September, US$4,265 in 1 to 27 October | A | TrustMRR monthly table. Calculated: November 2025 to January 2026 brought US$62.78; April to 27 October 2026 brought US$26,623, which is 99.7% of all revenue so far. September was 4.6% below August. | [1] |
| Revenue since launch, to 27 October 2026 | US$26,713 | A | TrustMRR all-time figure. Calculated: the monthly table sums to US$26,685.78, US$27.22 less; the table does not say where the difference comes from. | [1] |
| Users and downloads, 28 days to 1 October 2026 | 2,693 active customers, 2,351 new customers, about 2,400 downloads on iOS and Android | A | whatsthe.app, read from RevenueCat. Calculated: 211 subscriptions out of 2,693 active customers is 7.8%, the share that whatsthe.app calls subscription penetration. whatsthe.app also gives more than 500 downloads in total on Google Play. | [2] |
| Search traffic, 30 days to 27 October 2026 | 460,920 Google search impressions and 41,090 visitors | B | TrustMRR reads these from the site's Google Search Console and Google Analytics accounts, which it shows as connected. A search impression is one appearance of a page in Google's results. TrustMRR gives US$0.11 of revenue per visitor. | [1] |
| Pages on the website | 1,599 addresses in the sitemap: about 160 pages in each of nine languages besides English, 94 blog posts, 20 comparison pages, 36 pages for specific learner groups and 4 pages about other apps | C | practiceme.app/sitemap.xml, read 27 October 2026. Its last-modified dates fall mostly in April 2026 (454 pages), May (630) and June (300). Revenue rose in the same months; no source says that the pages caused the rise. | [8] |
| Marketing channels | App store optimization and SEO | C | Founder-written TrustMRR field. App store optimization means choosing the app's name, subtitle and keywords so that it appears in App Store searches; SEO (search engine optimization) means the same for Google. | [1] |
| Release on the App Store | 3 September 2025 | B | App Store version history (version 1.0 on 09/03/2025); whatsthe.app gives the same worldwide release date and TrustMRR the same founding date. | [3, 2, 1] |
| Founder and seller | Guillaume Sarfati, developer account in their own name | B | The App Store lists "Guillaume sarfati" as developer and seller, with one app on the account. whatsthe.app shows the same name and the X handle @tanisss_ with 185 followers. The copyright line reads "Copyright © 2026 CODELINE TEKNOLOJI". | [3, 4, 2] |
| Team at the October 2026 figure | 1 named person; headcount not published | C | No source names another person or gives a headcount. The website writes in the plural ("We have AI tutors specialized in different areas"). | [3, 6] |
| Funding | none reported | C | No source mentions investors. TrustMRR shows a profit margin of 68.0% for the last 30 days, a figure that is not among the metrics it says are verified. | [1] |
| App Store ratings | 1 rating in the US (5.0); 16 ratings in 10 countries with an average of 3.9 | B | US App Store page and whatsthe.app's count by country (Germany 3 ratings at 5.0, France 3 at 4.0, India 2 at 3.0, Turkey 2 at 2.5, China 1 at 1.0). The age rating is 4+. | [3, 2] |
| Time from store release to the October 2026 figure | 13 months | calc | Calculated: September 2025 to October 2026. | [3, 1] |
How to read this table: a grade A value was read from the app's payment accounts (RevenueCat, and Stripe for the web plan) by TrustMRR or by whatsthe.app. A grade B value is a record that Apple shows on the App Store, or traffic data that TrustMRR reads from Google's tools. A grade C value comes from the founder's own fields on TrustMRR or from the app's own website. A "calc" value was computed on this page from the rows above it.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Yearly Pro (App Store) | 5.00 | US$59.99 a year; unlimited practice after a 3-day free trial | |
| Weekly Pro (App Store) | US$9.99 a week; unlimited practice after a 3-day free trial | ||
| Practice Me Pro (web) | 19 | Unlimited conversations with every tutor, topic starters, memory across sessions; card payment through Stripe |
Calculated: US$59.99 a year is US$5.00 a month; US$9.99 a week is about US$43.29 a month. The web pricing page has a yearly option marked "SAVE 57 %" whose price loads by script and could not be read. The same page's description text says "Weekly $7.99 or yearly $59.99", while the App Store lists Weekly Pro at US$9.99. The pricing page says that "mobile app and web accounts are separate and not linked", so a subscription bought on one does not work on the other. Every new subscription starts with a 3-day free trial.
When did Practice Me launch, and how did it grow?
From an App Store release in September 2025, through seven months with almost no revenue, to about US$4,500 a month from July 2026. Each entry carries its grade.
- 20 August 2025CThe terms of service of PracticeMe carry this date as their last update. [7]
- 3 September 2025BVersion 1.0 of AI Speak English - Practice Me is released on the App Store under Guillaume Sarfati's developer account. [3, 2]
- November 2025 to January 2026AUS$62.78 of revenue in three months. [1]
- 20 and 22 January 2026BVersions 1.1.0 and 1.1.1 bring interface changes and chat history. [3]
- February and March 2026ANo revenue is listed for either month. About 50 website pages carry last-modified dates in these months. Version 1.1.2 on 28 March says that the tutor "now remembers context" and lists an "Upgraded AI model". [1, 8, 3]
- April 2026ARevenue rises to US$1,829. Versions 1.1.3 to 1.1.6, from 19 to 24 April, bring "Faster conversations and quicker agent replies" and bug fixes. 454 website pages carry April dates. [1, 3, 8]
- May and June 2026ARevenue of US$3,630 and US$3,232. 930 more website pages carry dates in these two months. [1, 8]
- July to September 2026ARevenue holds between US$4,461 and US$4,677 a month. [1]
- 1 October 2026Awhatsthe.app reads MRR of US$2,910, 211 active subscriptions and US$3,913 of revenue in 28 days from RevenueCat. [2]
- 27 October 2026ATrustMRR reads MRR of US$3,044 from 235 subscriptions, US$4,560 in 30 days and US$26,713 since launch. The app is not listed for sale on either site. [1, 2]
Why did Practice Me work?
Four causes, in the order they mattered. Each one is a reusable lesson.
Speaking practice without a human partner
The App Store page puts human tutors at US$15 to US$30 per 45 minutes. A Yearly Pro subscription costs US$59.99, and the tutor is available at any hour. The product sells one thing that classes and text-based apps give little of: time spent speaking and being answered.
Search pages in ten languages
The website has about 160 pages in each of nine languages besides English, plus blog posts, comparison pages and pages about other apps. Most of them carry last-modified dates from April to June 2026. TrustMRR shows 460,920 Google search impressions and 41,090 visitors in 30 days, read from the site's own Google accounts. The founder lists app store optimization and SEO as the marketing channels.
A free trial, then a weekly or yearly plan
Every plan starts with a 3-day free trial. On the App Store the choice is US$9.99 a week or US$59.99 a year, so a learner who stays more than about six weeks pays less with the yearly plan. whatsthe.app shows 211 paying subscriptions among 2,693 active customers on 1 October 2026.
Two payment routes
The app takes payments through the App Store, recorded in RevenueCat, and the website takes card payments through Stripe at US$19 a month. TrustMRR reads both accounts. The two routes are not linked, so a learner who pays on the web has a separate account from the one in the app.
How it was built
These details come from the founder-written fields of the TrustMRR profile (grade C), the App Store records (grade B) and the code of the website (grade B). [1, 3, 5, 2]
- A native iPhone and iPad app written in Swift, with Firebase as the backend and RevenueCat for subscriptions, according to the founder's TrustMRR fields.
- A web version with its own accounts, sold by card through Stripe; the website is built with Next.js, which its page code shows.
- An Android app on Google Play with more than 500 downloads.
- The tutor listens to the learner's audio and answers by voice; the App Store privacy section lists audio data as collected and linked to the user. No source names the speech or language models.
- The app's interface is offered in 33 languages, and the website in ten.
- Nine versions from September 2025 to April 2026; the last update before 27 October 2026 is version 1.1.6 of 24 April 2026.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
Who tried this before Practice Me, and why did they fail?
Four companies in the Loot Drop database sold English practice with human teachers, online or in classrooms, and failed. The values below are from that database and are not independently verified.
TutorGroup (iTutor)
The promiseLive English lessons with Western teachers for learners in Asia.
Why it diedChina's 2021 ban on foreign tutors removed most of its revenue, according to the database, while the cost of each new student and the turnover of tutors were already high.
Dada English
The promiseFixed pairs of Chinese children and North American teachers for paid online English sessions.
Why it diedChina's 2021 rules banned for-profit tutoring in school subjects, including English, and the database says all of its revenue came from that tutoring.
Wall Street English
The promisePremium English courses in learning centers in Chinese shopping malls, sold for career advancement.
Why it diedThe database says it paid for running costs with upfront course payments, and that the model failed when regulators restricted those payments and the pandemic closed the centers.
Teacher Finder
The promiseMatch language learners directly with tutors, without a language school in between.
Why it diedLarger tutor platforms with more money took the market, according to the database, and its simple automated setup could not keep up. Each lesson still needed a paid human tutor.
The failed startups come from Loot Drop, a free database of dead startups, retold here in our own words. Loot Drop describes its entries as AI-assisted summaries of public sources, so these cards carry no evidence grade.
What changed
TutorGroup, Dada English and Wall Street English each raised US$100 million or more to sell English practice with paid human teachers, either by video or in classrooms. Every hour of practice needed a teacher's hour, so the cost grew with every student, and all three depended on one country's rules, which changed in 2021. Teacher Finder tried the cheap version, a match between learners and tutors, and still had to pay a human for every lesson.
An AI tutor removes the teacher's hour. Practice Me sells unlimited spoken practice for US$59.99 a year on the App Store, and the only person named in the sources is its founder. Its sales also do not depend on one country: whatsthe.app counts ratings from ten countries, and the website is written in ten languages. The limit is that many other apps sell the same thing: the App Store page recommends ten similar apps, among them Speakerly, SpeakL and Gliglish.
Can one person copy Practice Me now?
Yes, but the product itself is easy to copy.
The app needed no funding and, as far as the sources show, no team. Revenue came in the same months in which the website grew to about 1,600 pages for search in ten languages; the sources do not prove that one caused the other. Many competitors sell the same AI speaking practice, so the search pages and the reviews are where one app differs from another.
Where it is still open: Spoken practice for other languages that many people learn for work, such as German, Spanish or Japanese, and for one exam in one market (for example the speaking part of one country's citizenship or university entry test).
- 1Pick one skill that learners rarely get to practice, here speaking, and build the app around a voice conversation with an AI tutor.
- 2Give the tutors clear roles (everyday talk, formal British English, travel, interviews, exam practice) so that a learner knows which one to choose.
- 3Write pages that answer what learners search for, in their own languages, and connect Google Search Console so that the traffic can be checked.
- 4Offer a short free trial and a yearly plan priced well below the weekly plan, and record payments in a service such as RevenueCat.
What could go wrong for Practice Me?
The risks are the mirror image of the causes above.
A selected rating on the website
The home page shows "5.0 on App Store", and its structured data, the hidden description that search engines read, gives a rating of "5.0" from a "ratingCount" of "3". The US App Store shows 1 rating of 5.0; whatsthe.app counts 16 ratings in 10 countries with an average of 3.9, including Germany with 3 ratings at 5.0. The rating shown is real but chosen from one store. The same structured data says "Learn English 5x faster with AI tutors"; no source supports that number.
Prices and accounts that do not agree
The pricing page's description says "Weekly $7.99 or yearly $59.99", while the App Store lists Weekly Pro at US$9.99 and the web page shows US$19 a month. The pricing page also says that "mobile app and web accounts are separate and not linked", so a learner who subscribes on the web cannot use that plan in the app. The home page's description says "Free on iOS & Android"; the App Store shows "Free · In‑App Purchases", and no source describes what a learner can do without a subscription after the 3-day trial.
Many apps sell the same practice
The App Store page recommends ten other English-learning apps, among them Speakerly, SpeakL, English Shadowing, TalkParty and Gliglish. Any of them can add the same tutors and voices, because none of them owns the models.
Traffic that depends on Google
The founder lists app store optimization and SEO as the marketing channels, and almost all revenue came after the website grew to about 1,600 pages in April to June 2026. If those pages lose their positions in Google's results, the source of new learners would shrink. The app itself has not been updated since 24 April 2026.
Unnamed models, an unexplained company and founder fields
No source names the speech or language models or their cost per conversation, so this page cannot say how the 68% profit margin on TrustMRR (not a verified metric) would change with model prices. The copyright holder CODELINE TEKNOLOJI appears only on the App Store page, and no source says how it relates to the founder.
Sources
The grade next to each source is the best grade that the source can support. "Read in full" means that the text of the page was downloaded and the numbers were checked in it.
- A TrustMRR markdown profile: Practiceme, read 27 October 2026Read the markdown profile in full. Revenue by day and month, total revenue, MRR and subscriptions are read from RevenueCat and Stripe through API keys, last synced 27 October 2026 (grade A). Traffic and search impressions come from the connected Google Analytics and Google Search Console accounts (grade B). The tech stack, marketing channels and profit margin are founder-written or not explained (grade C).
- A whatsthe.app listing: Practiceme, read 27 October 2026Read in full, including the data embedded in the page. Revenue, MRR, subscriptions, trials, customers and downloads are "verified through RevenueCat", last updated 1 October 2026; the founder's name and X handle are shown by the listing (grade C). The site's robots.txt allows this path.
- B Apple App Store: AI Speak English - Practice Me, US storefront, read 27 October 2026Read in full: description, rating, age rating, version history, privacy section, seller, copyright line, in-app purchases and similar apps.
- B Apple App Store: developer page of Guillaume sarfati, read 27 October 2026Read in full. The account lists one app, Practice Me.
- C Practice Me home page, read 27 October 2026Read in full, including the structured data in the page code. Source for the tutors, the platforms, the "5.0 on App Store" line and the Next.js page code. The site's robots.txt allows ClaudeBot.
- C Practice Me pricing page, read 27 October 2026Read in full. Source for the web price of US$19 a month, the 3-day free trial, Stripe as the web payment processor, the separate web and app accounts and the description text with weekly and yearly prices. The yearly web price loads by script and could not be read.
- C Practice Me terms of service, last updated 20 August 2025Read in full. Source for the date of the terms; they name no company and no age limit.
- C Practice Me sitemap, read 27 October 2026Read in full and counted with a script: 1,599 addresses with their languages, sections and last-modified dates.
- C Loot Drop database export, 1,749 failed startupsLocal export. AI-written summaries, not independently verified. Source for the failed predecessors.