Stringers
Mobile App · Android + iOS · written 11-6
Hark Pro · Hark Labs Inc.

Let AI order you a free pizza: how Hark Pro hit 330k installs, and why you can't copy it

What it is An AI assistant app that drives a browser on its own to book restaurants, buy groceries, process returns, and manage your calendar.

Why it blew up: 700M USD in revenue and a founder X account with a million views. Media ran coverage on launch day, the first 100k signups got the paid version free, 10-16 gave away 1M USD worth of free pizza and hit #1 on the iOS Lifestyle chart. After that, it kept paid install ads running on Google to hold its Android ranking.

DHow you could follow Don't build a general personal assistant. Pick one country and one type of site, build a narrow agent for it, and copy its interface that shows users what the AI is clicking, plus its small-scale acquisition stunt where the prize is having the AI do the task for you.

Grade D: it took off on huge funding, the founder's fame, and media relationships, which followers can't replicate. The category also has Meta Muse and OpenAI Dots, and Hark itself has only a fraction of Muse's scale. A few tactics are worth studying. Public / widely discussed
Data (updated daily) · launch 2026-07-20 · 11-07 installs 326,956 · launched 2026-09-30 · US store rating 2,716 · charted in 11-08 42 stores
Android · Total installs worldwide
150k200k250k300k350k10-2910-3111-0211-0511-07327k
Data 2026-10-29 → 2026-11-07 (10 days, 10 observations). Google's detail page only refreshes install counts every other day, so a flat day doesn't mean nobody installed.
Android · Ratings count (cumulative)
2.5k3k3.5k4k4.5k10-2910-3111-0211-0511-074.3k
Data 2026-10-29 → 2026-11-07 (10 days, 10 observations)
iOS · Top Grossing rank · by store
5010015020010-3111-0211-0411-0511-07186 Argentina189 UY
ArgentinaUY
Data 2026-10-31 → 2026-11-07 (8 days, 8 observations). Lower rank is higher; gaps mean it wasn't charted that day
iOS · Best category rank · US store
10015020010-2810-3111-0311-0511-08196
Data 2026-10-28 → 2026-11-08 (12 days, 12 observations). Best rank across each day's category charts
iOS · Number of stores charted (top 200 on any chart)
012310-2911-0111-0311-0511-071
Data 2026-10-29 → 2026-11-07 (10 days, 9 observations)

Hark Pro is an AI assistant app that drives a browser on its own to book restaurants, buy groceries, and process returns. It launched in October and got 330k installs on Google Play within a month. It isn't a hit from a small team. Behind it is Brett Adcock, founder of Figure AI, with a new AI lab and 700M dollars in funding. Its rise came from money, the founder's fame, and a 1M-dollar pizza giveaway. The tactics are worth dissecting, but the category itself isn't worth chasing.

A pizza on a Friday night

2026 Oct 16, Friday night. Hark's official account posted on X: download Hark Pro, tell it "order a free pizza," and it will order for you with a $30 DoorDash gift card, giving away $1M in total. The post got 1.06M views. Adcock's own repost got 1.78M views and 984 comments, and the deal-hunting groups on Facebook picked it up too.

The giveaway ended in under 9 hours. The company says it sent out nearly 20k pizzas. The next day Adcock posted, "usage hit unprecedented levels, and we broke a lot of things on our end." That night, Hark's own computer-use model, Handoff, ran 204k browser sessions. That day Hark Pro hit #1 in the US App Store Lifestyle chart and #58 overall (a screenshot Adcock shared).

There were real failures too. Gift cards weren't applied in time, the AI placed orders anyway, and hundreds of people were charged real money. One person's brother's credit card was charged $27.70; investor Sheel Mohnot said the next order took 20 minutes; Replika founder Eugenia Kuyda posted a string of failed screenshots and said she had "never worked this hard for anything." An Axios reporter tried several times, got a $30 gift card but no pizza, and later received a Google security warning about their Gmail. The company gave each person who was charged a $100 gift card (Axios 10-20; Food On Demand 11-04).

After the mishaps, Axios, Yahoo, and Food On Demand each wrote another round of coverage. Whether it worked or not, the stunt demonstrated the product itself: an AI logging into real websites, placing orders, and paying on your behalf.

What it is

The store page's first line reads: "Hark Pro is your life's operating system, thinking ahead for you and getting real, specific things done." The examples: it can log into OpenTable to book a table, check your calendar for free days, remember your favorite spots, note a friend's food allergy in the reservation, and generate a button to call an Uber that night (Google Play store page, 11-06).

It differs from ordinary chatbots in several ways: a small window in the chat shows in real time what it clicks on the web; the home screen proactively suggests things to do and supports Panels mini-apps; the interface was designed by a former Apple iPhone Air design lead; and it's marketed on "no ads, no selling your data." The free tier started with generous limits. The paid plan at $20/month gives 2x usage, and $100/month gives 10x, priced the same as Meta's Muse.

Where the volume comes from

DateGoogle Play cumulative installs
10-06Public launch (web, iOS, and Android at once)
10-08Officially, 24 hours after launch it had 143k DAU and 107k installs (both platforms combined)
10-29193,974
10-31237,129
11-03299,792
11-06326,956,7 days +133k

Data: Google Play install counts; 10-08 is the figure Adcock posted on X.

Four traffic sources stacked together:

The founder's X account. 10-05 Adcock teased "launching this week, the first 100k sign-ups get the paid plan," 1.14M views; the launch post got 10-07M views and 1.59 likes... 6,346. Afterward he posted a metric or a user testimonial almost every day.

The media all ran it the same day. 10-06 TechCrunch, Bloomberg, Gizmodo, Quartz and Fast Company published together, and a May 700M Series A at a 6B valuation, plus early coverage of a 2027-year hardware deal with AT&T, backed it up.

The pizza promo. See above.

Google paid acquisition, the reason Android keeps climbing. In Google's Ads Transparency Center, verified advertiser Hark Labs Inc's first creative started serving on 10-07, the day after launch. All the video creatives point to com.hark.android on Google Play, one creative running across 46 to 111 countries, a classic app install ad. Creative count went from 11-04 to 18 and then 11-05 to 32, and they're still adding. Installs rise in steps (10-31 +43k in a day, 11-03 +63k), which looks like spend released in waves (inferred).

The two sides make the point: the Android app with ads holds ranks 5 to 6 on the US Lifestyle free chart, and also charts in Brazil, Argentina and Nigeria. On iOS, the US Lifestyle rank has already slipped to 169 (App Store, 11-06). On pizza day it was #1 on iOS.

There's also creator reviews (on YouTube, Alex Finn used a link with a referral code) and batches of AI content-farm reviews. One 1-star Play review reads "forced to partner with an app, so I had to download this," and another asks "when is my money arriving?", which sounds like users from incentivized download channels. Only these two reviews back that up, unverified.

Why users stay, and why they complain

The praise clusters around three things: it actually gets things done, it acts on its own, and the interface is easy to use.

  • "I have severe ADHD. I gave Hark my task list for the week, it prioritized it, told me which items it could handle, then added reminders, put them on my calendar, and started researching." (5 stars, 10-26)
  • "Tried ChatGPT, Copilot, Gemini... I subscribed after just a few days. None of the other AIs got me to move from free to paid." (5 stars, 10-27)
  • "Useful to the point of making me uncomfortable. Like autocomplete for my life." (4 stars, 10-28)
  • "I think Hark Pro is about the same as Meta Muse." (5 stars, 10-31)

Late October they started tightening the free allowance, and the complaints followed:

  • "You suddenly cut daily usage by a lot." (1 stars, 10-29)
  • "If it's free, why do you want my credit card? What's the difference from asking for my social security number?" (1 stars, 10-28)
  • "Over half of what it says it did, it didn't actually do. Ask it to go back and check, and it finds errors every time; try to fix them and a whole month's allowance is gone in a few hours." (1 stars, 10-28)

(Google Play US reviews, 10-25 to 10-31, original in English, translation here.)

This is the standard playbook: give generously, build the habit, then tighten and push to paid. Revenue isn't public. Play's Lifestyle top-grossing charts in Argentina, Nigeria and Southeast Asia only place it between ranks 130 and 150, and iOS ranks #10 on Argentina's Lifestyle top grossing and #18 in Vietnam. Someone is paying, but not at much scale (inferred).

Who it's competing with

CompetitorsScaleVersus Hark
Muse from MetaGoogle Play 9M installs, 7 days +1.61M; hit #1 on the iOS US overall chartTied to Meta's social graph, with privacy controversy; Hark's Android scale is only about 3.6% of it
OpenAI DotsLives inside ChatGPT, which has 1.7B installs on PlayA feature, not a standalone app
ManusGoogle Play 11.5M installsLeans toward work output: video, slides, building apps
GensparkGoogle Play 4.97M installsLeans toward office work
InstinctInvite-only, runs over SMS and phone, no app to installReported valuation 10B, unverified

Data: Google Play, 11-06; CNBC 10-07.

The "personal AI agent that remembers things and does things for you online" route got crowded in 2026 September–October, with Meta, OpenAI and several star startups all piling in. Hark isn't the first, and it isn't the biggest. I haven't seen any small product that copies Hark, because a small team can't afford it: every "order it for me" runs on a cloud browser, and the more free users there are, the faster the burn.

If it were you

One line: don't follow. This is a race that starts at 700M and has Meta and OpenAI on stage, and Hark itself only got a fraction of what Muse got. What you can take away is three tactics, not this direction.

Who it suits: teams already building some kind of AI agent (even a very narrow one) who are stuck on how to make people understand what it can do on first use.

Who it doesn't suit: indie devs and small teams who want to build a "general-purpose personal AI assistant". You don't have the founder's million views, the same-day media, or the budget to fund a free tier, and you're up against a free version built into WhatsApp and ChatGPT.

How to do it: if you do it, pick a narrow scenario that Hark and the big players won't take seriously: one country, one language, one type of site (like a country's government appointment booking, or booking for a certain kind of local business), and make "actually gets it done" more reliable than a general agent. Copy two things from its UI: let users see what the AI clicked on the web page, and give the homepage a few "I can do this for you now" buttons so people aren't staring at an empty input box.

How to acquire users: use the pizza playbook, but a hundred times smaller. The prize is the thing your product does: have it order you a free coffee or grab a booking slot, so the demo and the acquisition happen in one go. Cap it and load-test first. Hark's failure came from the gift-card payment flow, not the AI. You don't have the budget to compensate hundreds of people $100, and one mistake means a pile of one-star reviews. Only after that, put a small budget into Google App install ads in one country and see whether your Android ranking holds.

When the window is: there's no window left for competing with Hark. The window for narrow-scenario agents depends on when Muse and Dots connect your type of site, and at the current pace that's measured in months (inferred).

What would show you're wrong: ① Hark publishes revenue, or a Play top-grossing chart puts it in the US Lifestyle top 50, showing standalone personal agent apps can make money on their own, not just big companies; ② a small team with no big funding gets a general personal agent past 100k installs on Play in a month; ③ the reverse: if Android falls off a cliff after Hark stops buying, as iOS has, then this month's volume was all bought, and the tactic itself needs discounting.

Data and sources

  • Installs, charts, ratings, reviews: Google Play, 10-29 to 11-06; App Store US, Brazil, Argentina, Vietnam, 10-28 to 11-06.
  • Store page copy and pricing: Google Play store listing, 11-06.
  • Official numbers and campaigns: posts by @adcock_brett and @hark_labs on X (10-05 to 10-17).
  • Pizza promo: Axios (2026-10-20), Food On Demand (2026-11-04).
  • Funding and product: TechCrunch (2026-05-21, 2026-10-06), The Rundown (2026-10-09), Fast Company.
  • Ads: Google Ads Transparency Center, advertiser Hark Labs Inc, 11-04 to 11-05.
  • Competitors: Google Play 11-06; CNBC (2026-10-07).
  • Where you see "estimated", "inferred" or "unverified", that's what it means.

Android store page ↗ iOS store page ↗

Research notes: competitor table, acquisition evidence, full research report
Competitors
NameWhereLaunchedScaleHow it differs from this one
Muse from MetaAndroid2026-09-08(Play 2026-08-31)Play 9M installs, 7 days +1.61M; hit #1 on the iOS US overall chartMuse is tied to Meta's social graph, with privacy controversy; Hark sells no ads and takes no data, shows the process as it runs, and emphasizes end-to-end completion
OpenAI DotsBig-company features2026-09-29Rides on ChatGPT (1.7B installs on Play)Dots is a feature inside ChatGPT; Hark is a standalone app with a proactive task homepage and Panels mini-apps
InstinctOther2026Invite-only, reported valuation 10B (unverified)Instinct is used via SMS and phone, with no app interface; Hark bets on a graphical interface
Manus AIAndroid2025-04Play 11.5M installsManus leans toward work output (video, slides, apps); Hark leans toward everyday life
Genspark AI WorkspaceAndroid2025-03Play 4.97M installsLeans toward an office workbench; Hark leans toward personal life and proactive recommendations
Where users come from
  • Confirmed founder traffic on X: @adcock_brett's post drew 1.59M views and 6,346 likes; the teaser "first 100k signups get the paid version free" drew 1.14M views
  • Confirmed concentrated media coverage: 10-06 TechCrunch, Bloomberg, Gizmodo, Quartz and Fast Company published on the same day; the May funding news
  • Confirmed cash-burning promo (free pizza): 10-17 gave away 1M USD of pizza; the official post got 1.06M views and nearly 20k were handed out; that day it ranked #1 on the US App Store Lifestyle chart; after the mess-up, coverage from Axios and Food On Demand brought another round of exposure
  • Verified Google App install and search ads: Google's Ads Transparency Center lists 33 creatives from Hark Labs Inc, first run 10-07. Video creatives land on com.hark.android and reach 46–111 countries. Creatives grew from 18 to 32 (googleads library, 11-05).
  • Confirmed KOLs and review accounts: Alex Finn reviewed it on YouTube using a link with a ref parameter; Andrew Wilkinson and others praised it on X, and the official account reposted the praise
  • Inferred offer walls or incentivized downloads: two Play reviews mention being "asked to download an app" and "money not arrived"; only circumstantial evidence

Why Hark Pro (com.hark.android) took off

Research date 2026-11-06. Sources: our googleplay / appstore / googleads / metaads / tiktokads libraries (agent-im server), live pulls from X (@adcock_brett, @hark_labs), and reporting from TechCrunch, Axios, Food On Demand, The Rundown, CNBC and others.

Bottom line

Hark Pro 不是小团队爆款。它是 Figure AI 创始人 Brett Adcock 的新 AI 实验室 Hark 的第一款产品:Adcock 个人先投 1 亿美元,公司 2026-05 又融了 7 亿美元 A 轮,估值 60 亿美元。产品赶在 Meta Muse、OpenAI Dots 抢起来的「个人 AI 代理」风口上线,靠三样东西把用户拉了进来:创始人在 X 上的流量、媒体集中报道、烧钱的活动(前 10 万个注册送付费版,10 月 17 日送 100 万美元免费披萨)。之后持续在 Google 投 App 安装广告,覆盖 100 多个国家,所以 Android 的安装一直在涨。差异化(自研的电脑操作模型 Handoff、聊天里直接看到它在浏览器里点什么、主动推荐要办的事、Panels 小应用、前苹果设计师做的界面、主打隐私)有真实的口碑支撑,但它跑出来靠的主要是资源和分发,不是产品本身的自然传播。

1. Basic data

  • Google Play: first listing date in the library 2026-07-20 (likely during internal or closed testing; public launch was 10-06). Installs went from 2026-10-29 to 193,974,11-06, up 326,956,7 over 132,982 days. Ratings count rose from 2,793 to 4,121, adding 140–196 a day (googleplay.app_day). On the US Lifestyle free chart it has held ranks 5–6, and on the US overall apps chart ranks 75–96; on Brazil's Lifestyle free chart ranks 16–31, Argentina 8–33, Nigeria 32–54. On top-grossing charts it sits at ranks 123–152 in Lifestyle in Argentina, Nigeria and Southeast Asia (googleplay.chart_rank). In-app purchase range $23–$230 per item.
  • iOS (id 6772362941): launched 2026-09-30. US rating count rose from 10-28 to 2,004 by 11-06 (now 2,623), adding 41–90 ratings a day. It now ranks #169 on the US Lifestyle free chart, #13 in Brazil (appstore.app_profile / app_day). On the day of the 10-17 pizza promo it hit #1 on the US App Store Lifestyle chart and #58 overall (Adcock's screenshots on X; Food On Demand covered it).
  • Official numbers (posted by Adcock on X): 24 hours after launch, 143,244 daily active users and 107,519 iOS/Android installs (2026-10-08); by 10-10, 562k users in total; Handoff browser sessions passed 10-15M 1.
  • Scale comparison: Meta's "Muse from Meta" (com.facebook.aura) went live on Google Play 2026-08-31. It now has 9M installs, has added 7M in 1.61 days, and reached #1 on the Play overall chart (googleplay.app_profile). Hark's Android footprint is roughly 3.6% of Muse's.

2. Competitors (same job: a personal AI agent with memory that operates a computer or the web to get things done for you)

ProductWhereLaunchedScaleHow Hark is different
Muse from MetaiOS / Android / WhatsApp / Mac2026-09-08Play: 9M installs, +1.61M in 7 days; iOS US overall #1 (CNBC 2026-10-07)Muse is wired into Meta's own social graph and has privacy controversy (Wired reported it building profiles of friends and family). Hark positions itself as "no ads, no selling your data," with a visible browser window for actions and end-to-end task completion.
OpenAI DotsFeatures inside ChatGPT2026-09-29Riding on ChatGPT's 1.7B Play installsDots is a feature inside ChatGPT. Hark is a standalone app whose home screen is a stream of proactive task suggestions, Panels mini-apps and custom interfaces.
Instinct(Spear Street Technology)SMS / phone / WhatsApp, no app to install, invite-only2026 yearsReported valuation of $10B (dapta.ai, unverified)Instinct has no app interface. Hark is betting on a graphical UI.
Manus AIiOS / Android / web2025-04Play 11.5M installsManus leans toward work output (making videos, slides, apps). Hark leans toward everyday chores (ordering food, groceries, returns).
Genspark AI WorkspaceAndroid / web2025-03Play 4.97M installsSame as above, leans toward office work
General-purpose agents like Perplexity / Comet and OpenClawWeb / open source2025–2026BigGeneral-purpose tools, not positioned as a "life operating system"
Dazzle(Marissa Mayer)App2026Not verifiedPhoto-centric

Takeaway: in this lane, from 9 2026 through October, Meta, OpenAI and several high-profile startups all crowded in. Hark is not the first, and not the biggest.

3. Where users came from (growth tactics)

  1. Founder traffic on X (confirmed): 10-05 Adcock teased "this week we're launching, the first 100k sign-ups get the paid tier" (1.14M views); 10-07 the launch post got 1.59M views and 6,346 likes; afterward he posted data and user feedback almost daily. @hark_labs's launch post got 180k views.
  2. Concentrated media coverage (confirmed): 10-06 On launch day TechCrunch, Bloomberg, Gizmodo, Quartz and Fast Company all published, and The Rundown followed. Earlier coverage included a $700M raise in May and a 2027-year hardware partnership with AT&T.
  3. Cash-burning promotions (confirmed):
    • The first 100k sign-ups get the paid tier free (10-05).
    • 10-17 launched "$1M free pizza": download the app, tell Hark "order a free pizza," and it orders for you using a $30 DoorDash gift card. The official post got 1.06M views and Adcock's post got 1.78M views and 984 comments; it spread in Facebook freebie groups too. The company says nearly 20k pizzas were given away and the promo ended after 9 hours, pushing it to #1 on the US App Store Lifestyle chart that day. Several hundred people were charged by accident. The company gave those people $100 gift cards, and Axios, Yahoo and LinkedIn News all wrote up the fiasco, which generated another round of exposure (Food On Demand 2026-11-04; Axios 2026-10-20). A Play review reads "pls remove your free pizza option."
  4. Google paid acquisition (confirmed, and the main reason Android keeps growing): In Google's Ads Transparency Center, advertiser Hark Labs Inc (verified) has 33 creatives, the first started serving on 2026-10-07, meaning they began buying traffic the day after launch. The video creatives land on com.hark.android, with app_install_share at 1.0, and each creative reaches 46–111 countries. Classic App install ads. Another 26 US text search ads. Creatives went from 11-04 to 18, and 11-05 rose to 32 creatives, 29 currently running, and they're still scaling (googleads library). This explains a few things: Android rankings and reviews in Brazil, Argentina, Nigeria, and Southeast Asia; Play's US Lifestyle chart holding at #5–6 while iOS has slipped to #169; and Play installs climbing in steps (10-31 +43k, 11-03 +63k).
  5. Meta / TikTok ads: none found in the library (metaads only tracks Pages on its watchlist, and tiktokads only captures ads served in Europe, so no results doesn't mean no spend).
  6. Creators and KOLs: YouTube has reviews from people like Alex Finn, who used a referral link, hark.com/?ref=alexfinn. On X, Andrew Wilkinson (@awilkinson) and others have praised it publicly, and the official account reposts positive user comments. AI marketing accounts like aiprofitboardroom and juliangoldie are also churning out batch reviews.
  7. Multi-platform rollout: web, iOS and Android all launched together. 10-30 came to Mac, and 11-01 supports iOS Live Activities. Adcock says 60 major features will ship within 12 days.
  8. Suspicious signals (unverified): One Play one-star review reads "Forced partnership with an app, which is why I had to download this app," and another says "taken a while to get the money into my account." These look like users from an offer wall or incentivized-install channel, but there's no other evidence. Also, a high share of 5-star reviews are short praise ("good," "super"), yet the US store has a batch of 1–2-star complaints from 10-27 and 10-28, mostly about tighter limits and credit card requirements, which doesn't look like deliberately steered praise.

4. Why users stay or pay

  • The free allowance was generous at first: "The free tier is generous right now" (Adcock, 10-08), and a user review says "just the free tier alone is changing my life." Limits tightened in late October, and 10-28 and 10-29 show one-star reviews saying "suddenly reduced daily usage limit," which suggests it's using the allowance to drive paid conversion. The paid tier at $20/month gives 2x usage, and $100/month gives 10x, matching Muse's pricing.
  • It actually gets things done: Fast Company tested it: it spotted from an email that the oil needed changing and helped book the appointment. TechCrunch reported it handled a California DMV vehicle registration renewal. Reviews from ADHD users and AI engineers praise it, and one user says "none of the other LLMs made me convert … into paid."
  • Proactive suggestions: Action Buttons and Panels on the home screen lower the barrier of "not knowing what to ask it to do," which TechCrunch identified as the hardest part of this kind of product.
  • Evidence of payment: On Play, it ranks #120–150 on the Lifestyle top-grossing chart in Argentina, Nigeria and Southeast Asia, and on iOS it's #10 on Argentina's Lifestyle top-grossing chart and #18 in Vietnam. Revenue is small and no revenue figures have been published.

5. Differentiation, and whether that's why it broke out

  • Differentiation: ① A proprietary computer-operating model, Handoff (the company says it beats both GPT and Opus, a claim of limited credibility), running up to 6 browsers at once; ② a small window in the chat that shows its web actions in real time, used to build trust; ③ a proactive home screen: suggested task buttons, self-built Panels mini-apps, and a background that changes with weather and time; ④ design led by Abidur Chowdhury, former iPhone Air design lead at Apple; ⑤ a privacy positioning of "no ads, no data selling," plus a marketing video of the CEO taking a lie-detector test; ⑥ the 2027-year hardware story (with AT&T, devices with built-in cellular).
  • Is it a product of spend, or real?: Only partly. Its early growth came mainly from resources: money, the founder's profile, media relationships, and the budget for a 1M-dollar pizza promo and global paid acquisition. Differentiation helped keep part of the bought users: reviews repeatedly mention the interface, proactiveness and getting things done, and some compare it to Meta's Muse. But its scale is a fraction of Muse's, and differentiation alone hasn't produced organic growth that can compete with big tech.

6. How many people know about it

Revenue is not public, but the company has been widely covered: TechCrunch, Bloomberg, Forbes, Axios and CNBC have all written about it, with a $700M raise at a $6B valuation, and the founder's X posts routinely hit millions of views. This is not an opportunity nobody knows about.

7. Copycats

I found no product specifically copying Hark. Comparable products come from big tech (Meta, OpenAI) or startups launching in the same period.

8. What to borrow

  • The promo design, "let the agent do one real task for you, with the prize being the task itself" (have the AI order you a free pizza), both drives acquisition and directly demonstrates the product. Even the botched execution earned media coverage.
  • Announcing "the first N×10k sign-ups get the paid tier" before launch creates scarcity and concentrates launch-day attention.
  • Google app install ads went live the day after launch and spread globally, so Android rankings can be sustained by paid spend. On iOS, rankings drop without it.
  • Showing the agent's actions in the interface builds trust, and proactive task suggestions on the home screen solve "not knowing what to ask it."
  • Generous free allowances build usage habits first, then tighter limits push conversion to paid.

9. Uncertain

  • Whether the Play launch date 07-20 was a closed beta or something else is unclear.
  • No data on Google spend, or on US campaigns on Meta and TikTok.
  • Whether offer walls or incentivized downloads were used rests only on two reviews.
  • Real retention and paid conversion: the company only says "the vast majority stayed."
  • The claim that Handoff "independently outperforms GPT and Opus" is unverified.
  • The exact plans behind the in-app purchase range of $23–$230 are unverified (likely the tax-inclusive prices of the $20 and $100 monthly plans, or annual prices).

Sources

This article was written by AI from public data and the data we collect daily. Numbers are sourced, and inferences are labeled.