Agentic Trading Demands Unified API
Kurush Dubash, CEO of Dome, on unified API for prediction markets
Agentic trading shows prediction markets are turning from destination apps into embedded execution rails. The important shift is that users increasingly do not open Kalshi or Polymarket and click through markets by hand. They type an instruction into a bot, follow a trader, or trigger a rule off live information, and software places the order for them. That behavior rewards APIs, low latency data, and smart order routing more than polished consumer interfaces alone.
-
The builder activity around Dome is concentrated in trading tools, not just market skins. The main buckets are analytics dashboards, copy trading products, and chat driven agents that buy contracts automatically, which suggests real demand for infrastructure that can read prices, track wallets, and execute orders fast across venues.
-
This looks similar to how copy trading worked at eToro. Users do not need to form every view themselves, they pick a trader, strategy, or prompt, then software mirrors the trade flow. In prediction markets, the extra twist is that on-chain activity at Polymarket is public, so builders can inspect wallets and reverse engineer tactics more easily.
-
The market structure makes automation especially powerful. Dome estimates that on Polymarket, fewer than 8,000 users out of roughly 1.4 million accounted for 80% of volume, with many of the heavy traders using arbitrage, copy trading, and bots. In a market shaped by concentrated power users, tools that compress reaction time can control outsized volume.
The next step is prediction markets being distributed through brokers, sportsbooks, wallets, and AI interfaces, with the winning infrastructure being the layer that normalizes markets and routes orders wherever liquidity is best. As more volume comes from software instead of manual clicking, the category will start to look more like electronic trading infrastructure and less like a standalone betting app.