How BottleCap AI is Bringing Efficient, Architecture-First Foundational LLMs to the World, Launched Pulse: Community News on IOS

BottleCap AI, making LLMs radically more efficient, led by Tomas Mikolov, Jaroslav Beck, and David Herel, has secured a $7.5M seed round led by 20VC alongside an incredible group of founders (Canva, ElevenLabs, Datadog, Huggingface, Lovable, Synthesia, Pigment, and many more).

Large language models continue to advance rapidly but remain inefficient, with progress largely driven by increasing compute rather than meaningful gains in reasoning. BottleCapAI focuses on architecture-first LLMs that prioritize efficiency by combining foundational research with real-world product development.

Lab breakthroughs are translated into practical applications designed for broad use. The first result is the launch of “Pulse: Community News” on iOS, built on CAP1, a proprietary foundational model.

Pulse began as an internal tool for aggregating AI and research news, enhanced with real-time community insights to surface meaningful signals beyond headlines. Built on the efficiency-focused CAP1 foundational model, Pulse marks the first in a series of applications developed on BottleCap’s proprietary technology.