A delivery rider turned their expense spreadsheet into an app: after Taiwan's delivery-worker law took effect, riders use it to demand money from platforms

Gonzo (工佐) is a delivery-rider bookkeeping app built by an indie developer in Taiwan. Each order: tap start and stop, and it works out the minimum pay that order should get under the platform delivery law effective Jul 21, 2026. Underpaid orders get flagged in red, you reconcile against the platform every two weeks, and one tap generates a complaint letter in the Ministry of Labor's format. It's small, with just over 10K downloads across both platforms and not much money yet. What's worth watching is how it timed a new law, earned union backing, and became #1 in a very small niche.

"Total income $3,741 + short payment $266.60"

On Sep 28, 2026, a delivery driver on Threads called 阿喵|外送员美食日记 posted his usual daily tally: 8 hours 59 minutes on the road, 33 orders, 111 km, $446 per hour; the last line read "Total: $3,741 + short pay $266.60 = $4,007.60". The post got 52k views.

"Short payment" is a new term. After the delivery-worker law took effect in July 21, platforms had to pay at least the statutory base fee per order, and any shortfall must be topped up every two weeks. But riders had no record of their own and didn't know how much the platforms were underpaying. The $266.60 in Ameow's post was calculated with Gonzo. In the replies, Gonzo's official account wrote: "This app is one of the most fully featured tools available right now... there is no referral code."

What it is

Developer Po-Cheng Yu, an individual who has only this one app. In June 20 it launched on Google Play first, and the first Instagram post said: "From a Google Sheets concept to a live Play Store release." The first line on the website reads: "An app written by Taiwanese delivery riders themselves." The iOS version launched in July 16, five days before the law took effect.

It builds a complete chain of evidence:

• Order timer: tap start on pickup and stop on delivery, and it calculates the minimum pay for that order using the law's formula (actual minutes ÷ 60 × minimum hourly rate of NT$196 × 1.25, with anything under NT$45 rounded up to NT$45), and flags red if the platform paid less. Stacked orders (multiple deliveries at once) are calculated separately for each order.
• Screenshot recognition: take a screenshot of the order screen, and the phone recognizes the amount, store name, and time on-device and starts the timer right away; customer personal information is automatically blacked out and never uploaded.
• Biweekly settlement: aligned to the platform's every-14-day top-up cycle, even matching the day-change times (Uber Eats at 4 a.m., foodpanda at midnight 12).
• GPS routes and PDF reports: every route and distance driven is logged, so it can be produced as evidence when needed.
• Complaint application: tick the short-paid orders and generate in one go a complaint letter in the Ministry of Labor's format, a disputed-order marking table, and attachments, with e-signature, packaged for sending.

No sign-up needed, and data lives only on the phone. This matters to riders: they're using these records to sue platforms and don't want any third party seeing their orders.

How it took off

Look at the numbers first. Google Play installs (third-party estimate): Jul 9 236, Jul 23 394, Aug 31 2,591, Oct 27 5,115, Nov 5 5,387. App Store rating in Nov: 6 ratings, average 4.88 (342), 24 added in the last 7 days. Since Oct 28 it has been on Taiwan's finance Top Grossing chart almost daily, peaking at #40 (10-31), and in Nov it was at #79 (6). The developer has said on Threads and Facebook that the iOS and Android versions each passed 5,000, over ten thousand combined, which matches these figures.

The curve has two steps, each tied to one event:

The first step: the law takes effect. Before it took effect, Android had 10 to 18 installs a day; in the two weeks after July 21 it jumped to 50 to 75 a day. The week before launch, Gonzo posted on Threads: "Tested with partner drivers — order timer, 7/21 law takes effect." Afterward, the union sued Uber Eats, the Ministry of Labor publicly ruled that Uber Eats' stacked-order pricing and some foodpanda orders were illegal, and riders for the first time had a reason to reconcile every order and file complaints with evidence.

The second step: union endorsement. In September 9, the National Delivery Industry Union posted on Facebook: "Two 'free' multi-platform delivery logging apps partnered with our union," with Gonzo ranked first, saying "Ministry of Labor complaint templates, generated in one click, signed online, packaged for mailing." The app can also send materials to the union in one tap. From late August to mid-September, Android added 90 to 134 installs a day, the highest stretch of the whole run.

The rest spreads slowly. Delivery-rider influencers like Ameow have long posted their "short payment" results; the official Threads account (826 followers), IG, and Facebook post a biweekly "reconciliation" round-up, for example for 10/12–10/25: foodpanda Gonzo calculated $1,254.5, actually paid $1,293; Uber Eats Gonzo calculated $1,362, actually paid $1,355.6. They use real numbers to prove their calculations are accurate, then spread into rider Facebook groups. The website breaks the law's text into Q&As, each question with its own URL that "can be shared directly with fellow riders," and the homepage shows a countdown to the "next top-up day." In late October it also entered the "Golden Source Award" indie developer contest and sought votes. No evidence of paid advertising was found (unverified).

After October, Android dropped back to 30 to 50 installs a day, and iOS ratings add 1 to 5 a day. This is a long tail that has passed its peak and is growing slowly through word of mouth, not a breakout.

Where the money comes from

The core features are free: timer, bookkeeping, complaint letters, routes and mileage. What's paid is unlimited screenshot recognition (free users get 3 a day), GPS route maps, and longer route history retention, offered as two subscription tiers, with the advanced tier free for the first 15 days; there is also a one-time donation to remove ads. Users who don't want to pay can unlock 24 hours by watching an ad. The in-app purchase prices listed on Google Play are $1.39–$12.99.

Taiwan's finance Top Grossing chart has a low bar, and being able to stay around ranks 40 to 85 for a long time shows steady small revenue, but the absolute amount is not large; estimated at tens of thousands of NT dollars per month (inference, unverified).

Ads were the biggest complaint. On Aug 16, a 3-star review said: "Watching an ad to unlock a fixed location is fine by me, but an ad after every finished order makes no sense." The developer later cut it to at most one full-screen ad a day. Most other reviews are along the lines of "Really useful, thank you so much to the developer."

Who it's competing with

This niche didn't exist before July 2026. Now three or four individuals and small companies are working on it, and Gonzo is the only one that covers both platforms and has real scale, and it also holds the union channel, the most trusted one. Taiwan's delivery riders are generally said to number a few hundred thousand (unverified), and that is the ceiling.

Few people know about it: it's well known in delivery unions, rider Facebook groups, and the delivery influencer circle on Threads; mainstream tech media and X haven't covered it, and revenue hasn't been disclosed.

If it were you

One-line take: Don't try to compete with it in Taiwan. This space is already held by it and the union, and the user group is small. What's worth learning is its playbook: when a new law suddenly gives a group of workers a right but no tool, whoever first builds the "calculate it and produce the evidence" tool gets recommended by the organization that represents those workers.

Who it suits: people who are themselves in a gig or precarious job (delivery, ride-hailing, housekeeping, courier, construction day labor), or who have people like that around them, and understand the industry rules and settlement details; solo developers who can build a small tool in great detail on their own and are willing to stay in the user community over the long term.

Who it doesn't suit: people who want a big business (the users are low-income and can only pay small amounts, and a region tops out at tens of thousands of users); people who don't understand local law or won't check formulas line by line (one wrong calculation and trust is gone); people who only know how to buy traffic and don't engage with unions and communities.

How to do it: Start where a new law has just taken effect or is about to, and sets a computable minimum pay or compensation. Examples: delivery riders' minimum pay, flight-delay compensation, overtime pay, deposit refunds. Build Gonzo's core: time each order or log it → calculate what's owed by the statutory formula → flag the gap → match the platform's payout cycle → generate a complaint in the official format. Keep the core free. Charge small amounts for convenience features like screenshot OCR and route maps. Store data only on the device.

How to get users: Make your first call to the union or advocacy group that represents this group and ask for a referral. Build "send it to them in one tap" into the product, which helps them too. Then find one or two practitioners who post their earnings on social media, and post regular "what we calculated vs. what the platform actually paid" reconciliation numbers. Turn the law into an FAQ site with one question per URL, so people can share each answer in their groups.

When the window is: It sits in the few weeks around the law taking effect. Gonzo listed five days before the law took effect, and two weeks after it went live, its daily installs had multiplied. Copycats arrived two months later. If you build this, have the product ready a month before the new law takes effect.

Signs you've misread it: ① The platforms make per-order pay breakdowns transparent and auto-top-up the difference, so riders no longer need a third-party ledger, and Gonzo's growth stalls. ② The law you picked already has an official tool from the union or the regulator. ③ Gonzo adds an English interface and "work regions." If it enters the market you wanted first, you're late.