Choose Your Interface

Lucrative markets hide behind legacy, expensive interfaces. AI lets us upend them.

The best store in Los Angeles has no website, no search bar, and no return policy. It’s the Silverlake Flea.

For more than twenty years, hundreds of vendors and thousands of shoppers have shown up every weekend. Each 10x10 booth holds a curated collection of vintage clothing, art, furniture, and electronics, sorted by item type, season, and occasionally price. Buyers comb the constantly revolving inventory for hidden gems. The allure of flea markets is the same phenomenon that has created the largest consumer marketplaces – affordable prices, quality goods, and trusted sellers.

Put another way, the Silverlake Flea is an expensive interface to a massive underlying market. Buyers scan tens of rows by foot, browse hundreds of pieces by hand, and settle on prices by haggling. They may find what they came for, but only at the cost of a Saturday. Some of us prefer the interface of local fleas, while others prefer a largely skeuomorphic experience – listing, browsing, and negotiating but from the comfort of a couch.

Today’s P2P marketplaces have digitized selection and search, but never reduced transaction costs. To participate on the sell side of the market, customers are expected to post listings, set prices, manage seller profiles, vet buyers, and negotiate through countless chats. The demand side of the market is less painful to participate in, but remains largely impersonal. The dominant players, an estimated $180B in annual GMV, have barely scratched the surface of proactive personalization. Each step is a cost, and those costs limit both demand and supply in the consumer goods market.

Lucrative markets are hiding behind legacy, expensive interfaces, and AI gives us the opportunity to tear the interface open. Remove the transaction costs and the wedge is clear: instant cash the moment you sign up. The work is ours. The find is yours.

Cost Reductions

A market exists only when the cost of participation is lower than the expected payout. For physical goods, that cost is the work of making a listing, sourcing demand, and handling logistics. The clothes in your closet, the furniture in your home, and the electronics at your desk are all items illegible to the market until you do the work.

Historically, the work was never worth it for the consumer. Over the past year, the cost to verify, specify, and market-make physical goods has reduced significantly.

1. Cost of specification (quality/use)

Cost of specification is the cost of defining an item and making it legible to the market.

Markets trade representations of reality, and each market defines reality differently (standards, listings, grading, SKUs, units). Language models make it nearly free to specify the long tail of unclaimed items and map them to relevant comparables. For example, a wrench can be:

• “Old wrench,” $10, available for pickup, FB Marketplace
• “Snap-on OEXM120B, 12mm,” $15, used, on eBay
• “Chrome vanadium steel, ~0.2 lb,” $0.02, sold to a steel mill

A single item holds three economic mappings. Before Tyn, an object had one identity: whatever the seller wrote. With Tyn, an agent routes all three to whichever markets clear the fastest, optimizing for liquidity on both sides of the market.

2. Cost of verification (condition/SKU)

Cost of verification is the cost of establishing that an item’s identity, condition, and ownership match its specification.

Tyn verifies an item’s identity, condition, and SKU from photo libraries, email receipts, computer vision, and credit profiling. More than 55% of consumer spending happens online, and the proliferation of POS for physical vendors has significantly increased the commonality of digital receipts. Shop, the third-largest commerce app, is built entirely on reading that inbox. For items with no digital record, a vision model now verifies condition from a single video for cents. Verification is cheaper because the evidence moved somewhere a model can read it.

3. Cost of market making (pricing and demand)

Cost of market making is the cost of sourcing a buyer and committing to a price.

Once an item is connected to Tyn, the agent prices it against comparable SKUs, historical sales data, and multi-platform search trends. Starting with a pricing reference, the agent then tailors the listing to each platform. Every item is also placed on Tyn’s orderbook, where vetted wholesalers place standing bids on verified items – driving “instant sale” offers for everyday users. A previously unquotable item now has a price and a bid. The long tail is economic for the first time.

Settlement moved down the cost curve too. Stablecoins turn a multi-day bank transfer into a near-instant payout. It isn’t in the product today, but we plan to embed compliance directly into the transaction, so credit rulings are enforced at the moment of transfer. The cost of specifying, verifying, and market-making physical goods has long exceeded the surplus of trading them, until now.

Market Maker Model

Tyn’s product flywheel is driven by faster payouts. Users sell with Tyn because it is the fastest, most reliable way to get paid for the items you own. Users buy through Tyn because it is the best place to find personalized, affordable goods. The stronger our instant-sell product, the higher quality our inventory, the more that users shop with Tyn. It is a self-reinforcing flywheel that drives both demand and supply.

What Comes Next

In an agentic economy, markets drive outcomes. OpenRouter aggregates models, quotes prices, and routes demand to the best available intelligence. Compute markets run the same model for hardware, aggregating and siphoning underutilized GPUs to meet demand as it appears. Commerce is one of the largest sectors of the economy and has yet to reconfigure around markets as the coordination layer. Cost reductions in verification, specification, and market-making are the beachhead. They make the long tail of owned goods legible enough to quote, clear, and eventually produce against.

Tyn’s edge in the market is understanding what the consumer already owns, what they are most likely to purchase next, and at what price.

To live in the future with us, head to Building an Institution.