Kalshi's new AI intern "Harrison" runs point on $5B World Cup betting bonanza ⚽

Crypto prediction market platform Kalshi has launched an internal AI agent called "Harrison" to help manage the surge in daily wagers the company is processing, including bets tied to the ongoing World Cup. The company disclosed the tool to Bloomberg, positioning it as a way to streamline operations across elections, sports, and awards markets. Kalshi co-founder Luana Lopes Lara said the agent is already performing daily tasks such as aggregating top news, analyzing competitor offerings, and making recommendations on what the platform should list next, including where to direct rewards for users who add liquidity. "We actually have an AI engineer in the markets team, where the AI is battle-testing the entire certification — finding out if you go in this direction, maybe there's a hole here, and all of that," Lopes Lara said. Lopes Lara also added that the agent was developed in-house.

The launch comes as Kalshi's trading volume has hit a record high above $5 billion, fueled heavily by World Cup bets. The Bloomberg report noted that the spike underscores how major sporting events are driving engagement on the platform and reshaping its user mix. The volume milestone arrives even as Kalshi and rival Polymarket face scrutiny from state regulators over claims that the platforms operate as unlicensed gambling venues. The U.S. CFTC has proposed new prediction market rules for these platforms and is also defending them in court against state-level challenges, a dual track that leaves the sector's regulatory perimeter unresolved in the near term.

Harrison's stated responsibilities go beyond market listings, with Kalshi framing the agent as a continuous quality-control mechanism that stress-tests the platform's own certification and surface area for new markets. The company says the system flags potential gaps before a contract goes live, a step that has become more sensitive as trading volume scales and regulatory attention rises. Lopes Lara described the agent as part of the markets team, embedded in the workflow rather than running as a separate product feature.