What happened
Kalshi Inc. developed an internal AI agent, named Harrison, to manage internal processes, specifically stress-testing prediction market contract wording. Built on Anthropic's Claude model, Harrison reviews contract language and evidence sources for millions of daily wagers across elections, sports, and awards, proactively identifying potential loopholes and ambiguities. The tool assists in designing and testing over 500 market templates, reducing the typical 1-2 hour human review process for new contracts.
Why it matters
Prediction market platforms can now automate complex contract validation, reducing operational load and potential disputes. This shifts contract design and review for legal and product teams from manual processes to AI-assisted analysis, enhancing market stability and execution efficiency. The move follows a broader trend of companies integrating AI agents for internal operations, such as Notion's recent launch of its AI agent platform, demonstrating a focus on automating core business workflows.
