BRENNEN

Industries / Technology

Marketplaces and gig platforms

Marketplaces and gig platforms are websites and apps that match buyers with sellers, or people who need a job done with freelancers who will do it, and that run the search, the trust, the payment and the rules for each match in return for a fee.

1995

In September 1995 Pierre Omidyar started, as its sole owner, the online auction service that became eBay: anyone could list a personal item, and strangers bid for it. Buyers paid $7.3 million for goods through it in 1996 and $95.3 million in 1997, counting only auctions that ended in a sale, by the company's 1998 prospectus.

sec.gov

38

Milestones achieved across 6 domains.

11

To watch for: announced, not yet confirmed outside.

12

Open deliverables. Nobody has shipped them.

In one minute

Where it started

In September 1995 Pierre Omidyar started the auction site that became eBay, and by 1997 buyers were paying $95.3 million a year for goods through it, by the company's count. Amazon let outside sellers offer goods on its own product pages from 2000, and Alibaba opened Taobao in China in 2003. Work followed goods: from 2005 Amazon's Mechanical Turk let businesses post small online tasks for anyone willing to do them, and from 2010 Fiverr sold freelance services at set prices.

Where it stands

Outside sellers made 61% of the units sold in Amazon's stores in the third quarter of 2026, and Amazon took in $46.5 billion in commissions and fees from them that quarter, by its own figures. The World Bank estimated in 2023 that 154 million to 435 million people, 4.4% to 12.5% of the world's labor force, do online gig work, and said that no reliable data source exists to count them. Rules are arriving: the EU fined Temu and AliExpress in 2026 over illegal products, EU countries must give platform workers rights over the software that manages them by 2 December 2026, and in June 2026 the International Labour Conference adopted Convention No. 193, a treaty on decent work in the platform economy.

What is open

Many open jobs are public tables built only from what platforms and lawmakers publish: what a sale costs a seller and when the money arrives, what decides search ranking, which laws ban fake reviews, and whether platforms' terms meet the new duties toward workers. Others are made-up data and test benches on which anyone can try detectors of side deals, scam chats and fake orders, pricing bots, marketplace payment cases, and AI-written letters that explain an account suspension. AI models that read legal text, browser agents and payment test modes, which move no real money, make each of these a job for one person or a small team.

Domains

AchievedWatch forOpen

Advances to build with

1 research note

Open in marketplaces and gig platforms

3 of 12 open deliverables

Open · Matching

Spot deals leaving the platform

An open, labeled set of at least 5,000 made-up chat threads between buyers and sellers and between clients and freelancers, in which some people try to move the deal or the payment elsewhere: phone numbers spelled out in words, handles split across messages, lookalike characters, offers of a discount to "pay me on another app." Every name, number and handle is invented. Publish it with baseline scores for a keyword list, a small trained classifier and two large language models.

Built with what exists

Claude, Gemini or an open-weights model such as Mistral Large 4 (Note: "A trillion open parameters") to draft threads; Label Studio for hand labels; scikit-learn; a small open model such as Liquid LFM3 for the trained baseline (Note: "One GPU to fine-tune"); Hugging Face Datasets to publish.

Credit · unclaimed

Open · Trust and reviews

Check review policies automatically

An open tool that reads a shop's or marketplace's public pages and reports, with quotes, whether it tells buyers how reviews are collected and checked, how it prevents fake reviews, and whether it bans or labels reviews given in return for a reward. These are the points EU authorities checked in 2022 and the CMA checked in 2025. The tool reads policy pages only and collects no reviews and nothing about reviewers.

Built with what exists

Browser Use 1.0 to open pages and follow links (Note: "Deleting 80% to start over"); Claude or Gemini to read each page and quote the evidence; Open Terms Archive for earlier versions of policies; Python.

Credit · unclaimed

Open · Holding and paying out money

A table of payout delays

A public table, checked every month, of when 30 marketplaces and freelance platforms release money to sellers and freelancers, read only from their published terms and help pages. For each platform and country it gives the days from payment or approval to money available, extra holds on new sellers, reserves, holds during disputes, the payout schedule and any fee for faster payout. Each cell links to the clause it comes from, and every change is logged.

Built with what exists

Open Terms Archive's open-source engine, which records each version of a web page of terms; Python and pandas; an AI model to draft each extraction, checked by a person; Browser Use, an open-source browser agent, for pages that need a full browser (Note: "Deleting 80% to start over").

Credit · unclaimed