Over the past 7 days, a new AI tool called Blanket has been quietly recommending event contracts to small businesses on Kalshi. But a deep dive into its architecture reveals a dangerous gap: the tool is a recommendation engine, not a hedge. And the contracts it suggests are binary, not proportional.
Context: The Regulated Prediction Market and Its Third-Party Catalyst
Kalshi is a CFTC-regulated prediction market—a platform where users trade contracts on binary outcomes: Will the temperature exceed 40°C in July? Will tariffs on Chinese goods increase by 10%? It's legally operating, with a license to offer event contracts. Blanket, built by a third-party developer, analyzes a small business's operational risks—weather, energy prices, tariffs, elections—and recommends specific Kalshi contracts to "hedge" them. It doesn't execute trades, handle funds, or process payments. It's a pure recommendation engine.
On paper, the value proposition is compelling: small businesses lack access to traditional derivatives or insurance for these niche risks. Blanket lowers the cognitive barrier. But the devil is in the technical details.
Core: The Technical and Regulatory Fault Lines
Let's start with the regulatory grey zone. Blanket deliberately avoids being a broker by not executing trades. But the act of recommending specific contracts could be construed as investment advice. Under the Investment Advisers Act of 1940, providing personalized recommendations for securities (or derivatives) requires registration. Event contracts, according to CFTC precedent, are not securities—but they are derivatives. If the CFTC decides that Blanket's AI constitutes "advice" on derivatives, the tool would need to register as a commodity trading advisor (CTA). That's a high bar. Kalshi, by outsourcing the AI to a third party, isolates itself from liability. But the third party is exposed.
What you see on-chain is not always what you get. Blanket's recommendation is just a signal. The actual trade happens on Kalshi's order book. And that's where the liquidity risk hits. The contracts Blanket recommends—like "Temperature in Phoenix > 45°C on July 15"—are long-tail events. They have low trading volume, wide spreads, and often no counterparty at all. In my 2020 analysis of the Uniswap liquidity crisis, I tracked gas spikes as liquidity providers fled. The same dynamic applies here: a small business owner buys a contract to hedge, but when the event occurs, they may find no one to sell to. The market is illiquid. The hedge becomes a lock-up.
But the bigger issue is basis risk. These contracts are binary: they pay out a fixed amount if the event occurs, zero otherwise. A small business exposed to a 10% increase in energy costs doesn't need a binary bet; they need a proportional payout. Blanket's AI might recommend the right direction, but the contract structure is fundamentally misaligned. The hedge doesn't cover partial losses. It's a all-or-nothing gamble, not a true hedge. Security is a promise; liquidity is the proof.
From my audit of the 0x protocol v2 in 2017, I learned that a seemingly simple function call can hide reentrancy vulnerabilities. Blanket's AI is similar—it looks straightforward, but the underlying assumptions are fragile. The AI model uses natural language processing to parse news, weather forecasts, and tariff announcements. But tail events—like a sudden tariff escalation—are precisely the moments when the model fails. The training data likely doesn't cover extreme scenarios. The result: the AI recommends the wrong contract at the wrong time.
Contrarian: The Unreported Angle—Blanket Is a Trojan Horse for Kalshi's Growth
The mainstream narrative is that Blanket helps small businesses. The contrarian view: Blanket is a customer acquisition tool for Kalshi, disguised as a risk management service. Kalshi gets transaction fees, but bears none of the AI development cost or liability. The third-party developer gets subscription revenue or referral fees. The small business gets a binary bet that may not hedge anything. The real winner is Kalshi, which builds network effects as more users trade event contracts.
But there's a deeper risk: if Blanket successfully educates small businesses to use prediction markets, it could attract regulatory scrutiny that kills the entire model. The CFTC has long been wary of political event contracts. Blanket's recommendation of election contracts could be seen as encouraging retail speculation. If the CFTC bans political contracts, Kalshi loses a major revenue stream. Blanket becomes a liability.
Takeaway: The Next Watch
The next 12 months will determine whether Blanket becomes a blueprint for regulated prediction market hedging or a cautionary tale of regulatory arbitrage. Watch for CFTC guidance on whether AI recommendations constitute investment advice. If the answer is yes, Blanket's model collapses. If no, expect a wave of copycats—and a wave of small businesses that learn the hard way that binary contracts don't hedge proportional risks. Volatility isn't the market; it's the market's language.
