Modern office workspace with a computer monitor
AI

Crypto.com says ai.com agent is still in stealth mode eight months after Super Bowl ad

The site still displays that AI agent generation is queued due to high demand, and no public launch date was provided.

By Emma Carter6 min read

Crypto.com confirmed its ai.com personal AI agent platform remains “building in stealth mode” as of Sept. 30, nearly eight months after a $15 million Super Bowl ad drove traffic that crashed the site within minutes. The ai.com website still tells users that “AI agent generation is queued due to high demand,” even as rival venues roll out narrower AI trading assistants.

Key Takeaways

  • Ai.com promoted its AI platform in a $15 million Super Bowl ad in February, then crashed within minutes as users tried to reserve handles for personal agents.
  • As of Sept. 30, 2026, the ai.com site still displays “AI agent generation is queued due to high demand,” indicating the product remains effectively unavailable.
  • Crypto.com confirmed the platform is “building in stealth mode” and said it will share more details when it “fully launch[es] to the public,” without offering a date.
  • While ai.com stays queued, competitors have shipped trading- and portfolio-oriented AI assistants, including Robinhood’s in-app automated trading agents, Kraken’s AI financial tools, and Bitget’s GetAgent.

Eight Months After the Super Bowl Splash, ai.com Still Shows a Queue

Ai.com, described as a business line of Crypto.com, is still not publicly live nearly eight months after it used one of the most expensive marketing stages in the industry to introduce the product. On Sept. 30, 2026, the ai.com landing experience continued to show users a status message: “AI agent generation is queued due to high demand.”

Crypto.com confirmed the platform remains under construction. A spokesperson said, “ai.com continues building in stealth mode. We are happy to provide more details when we fully launch to the public,” and did not provide a launch date, access criteria, or a timeline for when queued users might be able to generate agents.

That status sits awkwardly against the February rollout. Ai.com ran a $15 million Super Bowl ad in the fourth quarter promoting the launch of its AI platform, and the website crashed within minutes as users attempted to reserve handles for personal AI agents. The current on-site message does not disclose the size of the queue, whether any users have received agents since February, or whether access is limited to a closed test.

What ai.com Promised: Personal Agents Across Apps, Messages, and Projects

The February pitch was broad, and it was framed less like a single exchange feature and more like a general-purpose “personal agent” layer. Ai.com’s launch announcement described agents that could organize work, send messages, operate across applications, and build projects.

The same announcement also claimed the agents could develop missing capabilities themselves and share those improvements across the network, a promise that implies a system that learns from user activity and propagates new tools or workflows beyond a single account. That is a materially different ambition than the “assistant” features exchanges have historically shipped, which tend to be bounded to market data, portfolio views, and customer support.

Two terms matter for how traders should interpret the current state. An AI agent, in this context, is software that can take actions on a user’s behalf based on instructions and context, rather than only answering questions. “Stealth mode” is a development posture where a product is being built privately with limited public details and no broad release, which is consistent with the lack of a public spec sheet, pricing, supported integrations, or an execution policy for what the agent can do inside financial accounts.

What remains unspecified is the part that determines whether this becomes trader tooling or just a brand extension: whether ai.com agents will be able to execute trades directly, which exchanges or apps they will connect to, and what constraints will govern automation.

Rivals Ship Trading-Focused Assistants While Crypto.com Builds Offstage

While ai.com remains queued, the competitive baseline for “AI on a trading venue” is moving toward narrower, shipping products that focus on trading and portfolio workflows. On Tuesday relative to the Sept. 30 publication date, Robinhood announced new in-app AI agents that can perform automated trading. Kraken added AI-powered financial tools in July aimed at helping users identify investment opportunities and trades based on their financial goals. Bitget launched an AI crypto trading assistant called GetAgent in June.

Those releases are not the same product category as ai.com’s February narrative, but they do set expectations for what an exchange-delivered assistant should do first: surface opportunities, translate goals into actions, and, increasingly, automate execution in a controlled way.

Crypto.com has also signaled it is willing to deploy AI in targeted, execution-adjacent form even as ai.com stays in stealth. In March, the company introduced an integration with OpenClaw that allows customers to deploy a personal AI trading agent that handles trade execution. That same month, Crypto.com announced a roughly 12% workforce reduction, citing company-wide AI integration, and said the cuts were part of plans to prioritize resources around key growth areas.

On governance posture, Crypto.com obtained ISO/IEC 42001:2023 certification in February, an international standard for AI management systems covering governance and responsible development and deployment. Certification does not answer product questions like scope or launch timing, but it does indicate the company has invested in process around how AI systems are managed, which becomes more relevant if ai.com agents are eventually permitted to take actions across accounts and applications.

The Crypto.com ai.com agent still unlaunched Milestones Ahead

The next meaningful update is not another “still building” confirmation, but a public launch package that pins down mechanics. That includes a release date or window, eligibility, pricing, and which exchanges or apps the agent can connect to, along with a clear statement on whether agents can execute trades directly or only generate suggestions.

On the product surface, a change in ai.com’s on-site status from “queued due to high demand” to live agent creation would be the first observable shift. An explicit waitlist size, estimated time to access, or staged rollout plan would also clarify whether the queue is a temporary scaling issue or a placeholder while the product remains closed.

Competitive pressure is likely to come from feature creep on rival assistants. If Robinhood, Kraken, or Bitget expand from assistant-style features into direct execution, risk controls, or broader portfolio automation, the bar for what traders expect from “AI agents” on a venue will move again.

Crypto.com’s own governance signals could also become part of the launch story. The company’s ISO/IEC 42001:2023 certification is a concrete artifact it can point to, but traders will still need to see how that translates into user-facing controls, permissions, and guardrails once consumer-facing agent capabilities are actually shipped.

How Traders Should Read the Delay: Hype-to-Delivery Risk and the Automation Arms Race

The filing-equivalent detail here is the on-site state, not the marketing memory. A $15 million Super Bowl push that ends with “AI agent generation is queued due to high demand” eight months later is a clean execution risk signal, because it means the flagship “personal agent” narrative has not yet turned into a tool traders can evaluate, stress-test, or build routines around.

The threshold that matters is whether Crypto.com converts ai.com from a brand promise into a bounded, permissioned product with clear execution rules, because until launch mechanics and access are explicit, rivals’ narrower assistants will keep defining what “AI on an exchange” practically means for automation and workflow.

Sources