MoonPay has acquired AI research firm Dawn Labs and launched Dawn CLI, an AI trading tool announced around May 11–12, 2026. The move pushes MoonPay beyond payments and wallets into infrastructure for autonomous AI agents that can design, test and execute strategies on prediction markets.
The acquisition also reframes MoonPay’s role in market infrastructure. By combining payments, wallets and AI-assisted execution, the company is building a non-custodial stack for agentic trading, where automated systems can interact with markets without handing custody of user funds to a centralized provider.
BREAKING: MoonPay has acquired Dawn Labs pic.twitter.com/cC0lA553Ew
— MoonPay 🟣 (@moonpay) May 11, 2026
Dawn CLI Turns Strategy Prompts Into Executable Code
MoonPay integrated Dawn Labs and appointed its founder, Neeraj Prasad, as Chief Engineer of MoonPay Labs. The new Dawn CLI product is described as a command-line AI trading copilot that converts plain-English strategy descriptions into executable code.
The tool allows users to simulate and backtest strategies before moving them into live execution. That workflow makes testing and review part of the trading process, reducing the risk that an AI-generated strategy moves directly from prompt to market without validation.
Initial targets for Dawn CLI include prediction markets such as Polymarket and Kalshi. Those venues could see more automated and strategy-driven order flow if agentic trading tools gain traction among advanced users and developers.
MoonPay positioned the product inside a broader agent infrastructure stack. That includes MoonPay Agents for non-custodial agent infrastructure, MoonAgents Card for stablecoin spending, and work on the Open Wallet Standard to support locally generated wallets controlled by users.
Non-Custodial Controls Shape the Risk Model
MoonPay emphasized a non-custodial wallet design as a core safeguard. Under the model, wallets are created locally on the user’s device, so MoonPay and related agent services do not hold user funds, reducing direct custody and counterparty exposure.
The second safeguard is auditable strategy code. AI-generated trading logic is rendered in human-reviewable form before deployment, giving users and operators a way to inspect the strategy before it reaches live markets.
The third control is granular policy enforcement. Users can set limits on capital allocation, permitted markets and position sizing, so agent behavior remains bounded by explicit trading rules rather than open-ended model output.
These controls are designed to address core risks in autonomous trading, including unintended executions, excessive transaction activity and algorithmic hallucinations. Simulation and backtesting add an additional pre-trade filter before agents are allowed to operate in supported venues.
Non-custodial architecture may reduce custody exposure, but access controls, audit trails and policy enforcement become more important as automated agents interact with markets.
Prediction-market venues will also need to adapt operationally. Higher agent-driven activity could increase pressure on API throughput, rate limits, order surveillance and incident response for automated trading behavior.
MoonPay’s acquisition of Dawn Labs signals a broader move toward toolchains that connect payments, wallets and algorithmic execution. The key market question is whether transparent code, local custody and user-defined limits can make AI trading agents scalable without creating new operational risk.