The graveyard of AI hardware is crowded with failed pins, pendants, and glasses. Yet, one category is quietly thriving: the dedicated meeting recorder. While standalone AI gadgets struggle to find a purpose, Pocket has found a rhythm. The company just raised $11 million to prove that hardware still matters.

Accel led the round, joined by Y Combinator and ElevenLabs CEO Mati Staniszewski. It is a significant vote of confidence in a market that many assumed would be swallowed by smartphone apps. Pocket isn't trying to replace your phone. It is trying to stick to it.

The Hardware Play

Pocket sells a $129 credit card-shaped puck. It attaches magnetically to the back of your smartphone. The premise is simple. You press a button, and it records. It transcribes. It captures the room.

Unlike software-only tools like Otter or Fireflies, Pocket is designed for the physical world. It targets lawyers, doctors, and construction workers who move between meetings. These users need context that a Zoom call doesn't provide. They need offline access. They need reliability.

"We thought every meeting notetaker was built for online conversations, but nothing was geared towards real-life talk," co-founder Akshay Narisetti said. "AI really needs a lot of context to work better for us, and a lot of that context exists offline."

Betting on Momentum

The numbers suggest the strategy is working. Pocket claims to have sold 130,000 units since its launch last year. That is scale. It is also a clear signal that users are willing to pay for a dedicated device rather than relying on a phone app that drains battery and interrupts workflows.

While the hardware is the hook, the software is the engine. The base device offers free transcription. For $200 a year, users unlock advanced features: AI summaries, daily highlights, and deep file integrations. It is a classic razor-and-blade model, updated for the age of large language models.

The Enterprise Pivot

Hardware is just the entry point. Pocket is now pushing into the enterprise. It offers webhook support and integrations with tools like Obsidian, Claude, and Cursor. It even includes a model context protocol (MCP) server to link its AI assistant to external databases.

This is where the real battle lies. Pocket isn't just competing with other pucks like Plaud or Mobvoi. It is fighting for space against established software giants like Zoom and Read AI. These companies have deep pockets and massive user bases. They are all racing to automate the same tasks: drafting emails, updating CRMs, and assigning action items.

Key Takeaways

  • Hardware as a Service: Pocket has sold 130,000 units, proving that users want dedicated devices for high-stakes recording.
  • Offline Advantage: By focusing on real-life meetings rather than digital calls, Pocket captures context that software-only tools often miss.
  • Enterprise Integration: The company is moving beyond simple transcription, building deep hooks into professional workflows like CRMs and coding environments.

What This Means for Users

The market for AI note-taking is bifurcating. On one side, you have software that lives in your browser. On the other, you have hardware that lives in your pocket. If you spend your day in back-to-back Zoom calls, the software is likely enough. If your work happens in the field, the hardware is starting to look like a necessity.

Pocket’s next challenge is software velocity. They must ship integrations faster than their competitors. They must prove that their AI can handle the nuance of real-world conversation better than a generic transcription engine. The funding is in the bank. Now, they have to execute.