A French money-saving show made an AI that really orders your groceries for you double its installs in one week

Jowzi is the "little sister" of Jow, a French recipe app: you say "5 quick dinners this week, under 70 euros," and it plans the menu, fills the cart, and orders and pays through Carrefour or Intermarché for you. On Google Play it sits at #7 on the France Food & Drink free chart, with 77k cumulative installs, most of which came after the October 11 TV show. Below: why it took off, what users are complaining about, why a small team can hardly follow this opportunity, and a few tactics worth copying.

Sunday night, 9:10 on Capital

2026 October 11, a Sunday. The flagship finance show "Capital" on M6 aired an episode titled "Mission pouvoir d'achat: récupérez 400 euros chaque mois" (purchasing power mission: get back 400 euros every month). One segment featured a woman named Cindy. The show's Facebook caption read: "With the Jowzi app, Cindy planned her week's menu and placed her order in a few minutes."

That night, a batch of new reviews appeared on Google Play, clustered around 12 4 a.m. Beijing time, which is right after the show ended on Sunday evening in France at 10 p.m. Days later someone wrote: "I'm glad I found it through TV, it's fast, and you can plan a week of meals for little money." Others gave one star while saying "it's been updated several times, appeared on M6, and still doesn't deliver to your door."

Installs jumped accordingly. Third-party estimates put October 14 at about 15.6k, October 15 at 34.5k, with close to 19k arriving in a single day (Play's counts refresh in batches, so the exact day is uncertain); October 21 48.5k, November 6 77.3k (Google Play installs, 11-06). Jow itself posted on LinkedIn: "Vu à la télé" (seen on TV).

What it actually does

The store page's first paragraph names ChatGPT as the rival:

ChatGPT can give you recipes and cooking advice, but it can't buy your Monday groceries for you. Jowzi can.

It comes down to three things:

• Turn one sentence into a menu and a cart: type or speak "give me 5 quick dishes under 70 euros" or "I need toilet paper, two cans of sardines, parmesan...", and it generates the menu and puts the ingredients and household items into the cart of your usual supermarket.
• Order for you: connect your supermarket website account, and it handles product selection, payment, and choosing a pickup or delivery slot, without the supermarket needing to open a dedicated API.
• Track it for the whole family: a "Ping la famille" button sends the list to a family WhatsApp group, where anyone can say what they're short on, and it's added to the cart automatically; frequently bought items are sorted by room, time slot, and family member in a feature called "Magic List."

The app is free with no in-app purchases. The money comes from supermarkets and brands: Jow has long taken a commission from supermarkets on orders and also sells "recommendation slots" to brands, and trade press reports this makes up 20%–30% of Jow's revenue. Jowzi's pitch to retailers is blunt (per LSA's reporting): its recommendations "let retailers prioritize certain products, letting e-commerce earn healthy gross margins again." Jow says recommendations can account for 80% of the cart, and that 80% is the slot it can sell.

It's a new product from a 7-year-old company

Jowzi didn't come out of nowhere. Its developer Jow is a Paris company that launched in 2018; the main app has 6.37M installs on Google Play and sits steadily near #2 on the France Food & Drink top-grossing chart; the App Store France has 48.6k ratings. Jowzi's store description opens with "Jow's little sister, Jow has 10M users," and the jow.fr homepage links straight to Jowzi.

So its traction stacked up like this:

The 10k installs before TV came from the parent brand: existing users, the website, an Instagram account with 637k followers, and one round of industry PR. The TV moment worked because the topic was right: inflation in France over the past two years has made "purchasing power" a national conversation, and "AI helps you save 400 euros a month" fit exactly the story the show wanted to tell.

Also, the bar to reach the top of the France Food & Drink free chart is not high: about 12k new installs a week is enough to land at #7. One exposure pushed it up there, and the chart itself then brought in organic volume.

Almost nobody discusses it on X; the official TikTok account @jowzi.ai has only 700 followers, with videos getting a few hundred views. Nearly all its traffic comes from French TV, the parent brand and Instagram.

What users are complaining about

There are plenty of positive reviews, but nearly all the negative ones say the same thing: too few supermarkets.

"Where are Auchan and Leclerc? When will they be here? I only have Intermarché and Carrefour, why?" (2 stars, 10-20)

"It forces you to pick one supermarket right away. I don't shop there, I don't have an account, and I don't want to open one." (2 stars, 10-12)

"Pretty much like the Jow I already use, but way fewer supermarkets to choose from than Jow. Disappointed." (3 stars, 10-21)

Leclerc gets named most often, and Leclerc happens to be one of France's largest supermarket groups, which built its own voice-to-cart feature with Google. The main Jow app already connects Auchan, Super U and other major chains, while Jowzi currently connects only 3, so users naturally compare.

Two other complaints: bugs ("I tried removing ingredients I already have at home and the app crashes every time", "I set 2 people and it planned meals for 0"), and confusion over how it differs from Jow ("What's the difference between Jow and Jowzi?", "Other than using AI, it doesn't offer anything more than Jow"). Users outside France also say they installed it and couldn't use it.

One more detail: on the evening of Oct 5, over roughly 3 hours, 7 5-star reviews came in back to back, with lines like "Jow's new experience is really useful, recommended." The tone suggests they were solicited together (inference).

Growth is slowing too. After the TV segment it was adding thousands a day; now it's a few hundred to just over a thousand a day (between Nov 4 and 6 it added about 1,000). Order conversion and retention aren't visible from outside.

Who it's competing with

This table shows one thing: in France, the layer of "weekly meal plans by budget + price lists" already has newcomers crowding in (Piatto and Yuma both launched in October and charted in the top 15 of their category). The "place an order" layer only has Jow and the supermarkets themselves.

Why small teams can't follow

Jowzi's real foundation isn't AI. Jow spent 7 years connecting to the e-commerce systems of 10k+ stores across France and the US, so it can pick items, pay and choose delivery slots inside the user's supermarket account. The LLM layer is something anyone can plug in; this integration layer can't be copied quickly, nor bought.

Its distribution also rests on things you can't get: a parent app with 6.37M installs, familiar faces in trade media, and a brand story picked up by the TV show Capital. A new app doesn't get onto that episode.

The parts it hasn't done well yet (more supermarkets, Leclerc) are behind the same wall, and they're the parts Jow itself is most likely to fill in first.

If it were you

C: worth learning from, but don't go head-on. A "grocery AI that orders for you" is a game Jow and the supermarkets play in France. The moat is integration and the parent brand, not the model.

• Who it suits: teams that already have retail or e-commerce integration capability (who have built interfaces or proxy ordering for a national supermarket, fresh food platform or pharmacy chain); or people with channel relationships in a country where online grocery is concentrated among two or three big chains.
• Who it doesn't suit: indie devs who only wrap an LLM with no retail integration; anyone wanting to do weekly menus in France (Piatto and Yuma entered in October, and Jow itself offers it for free).
• How to do it: don't touch "ordering"; copy three moves. First, make ChatGPT your rival in the store's opening line: "it can talk, I can get it done." Second, use a WhatsApp group as the entry point so others at home can add items without installing the app. Third, make it free for users and sell "being recommended by AI" to brands. If you do build ordering, pick a market where one chain covers most users, and where there's no Instacart and no supermarket agent yet.
• How to acquire users: copy its TV angle, not the TV appearance itself. Package the product as "how much you save per month" and pitch local media and money or family shows on saving-money features; on Instagram, use creator-style short videos like "I bought a week's groceries in 5 minutes" with a first-order discount code.
• When to move before it's too late: Carrefour already has its own agent, Instacart is in ChatGPT, and supermarkets and large models will make "assistants that can place orders" standard within 2027 years. To get free coverage with the fresh story of "AI that shops for you," the window is probably before the first half of next year (inference).
• What would prove this read wrong: ① Jowzi picked up Leclerc and Super U before early 2027, and its daily installs went back to several thousand after the TV ads ended. That means this isn't just the parent brand getting one extra push. The product itself has pull, so it's worth re-evaluating. ② A new app with no parent brand reaches the top 10 of a category chart using the same "log into the supermarket account for the user" approach. That would show the integration isn't that hard and a small team has a shot. ③ The reverse: Jowzi gets merged back into Jow or stops updating in 2027. If even Jow couldn't make a standalone agent app work, this direction is one to avoid.