BottleCap AI

What BottleCap AI does

BottleCap AI is a Prague-based AI startup founded in 2025 that develops efficiency-first foundational large language models through novel architectures that explicitly move beyond the transformer paradigm. Its core technology is the CAP1 foundational model, which the company describes as architecture-first and designed to reason better per unit of compute, with a stated roadmap toward 100x efficiency gains over current LLMs. The company has operationalized a proprietary training algorithm that reduces LLM training costs by up to 50%, and pairs in-house foundational research with immediate productization rather than licensing its models.

The company commercializes through in-house applications rather than direct model sales. Its first product is Pulse: Community News, an AI-powered news aggregation app distributed globally via the iOS App Store and Google Play that surfaces real-time community sentiment on current events, running entirely on the CAP1 model. The company plans to layer enterprise applications on top of CAP1 with subscription-based monetization, in addition to consumer freemium mechanics, but currently discloses no public pricing. Pulse is described as the first of multiple planned applications on the CAP1 substrate.

BottleCap was initially bootstrapped by co-founders Tomas Mikolov (Chief Scientific Officer, creator of word2vec and the RNNLM toolkit, with prior research roles at Facebook, Google, and Microsoft), Jaroslav Beck (CEO, co-creator of Beat Saber, which sold 10M+ copies and was acquired by Meta in 2019), and David Herel (Senior Scientist and Head of Development, with publications at ICLR and ECAI). In January 2026 the company closed a $7.5M seed round led by 20VC with participation from founders of Canva, ElevenLabs, Datadog, Hugging Face, Lovable, Synthesia, Pigment, Supercell, and former Stripe CTO David Singleton. Headcount is stated as a small team of 1–10, operating full-time onsite in Prague, with active hiring for NLP researchers and machine learning engineers.

Name

BottleCap AI

Legal name

BottleCap SE

Website

https://bottlecapai.com

Company type

Private

Founded year

2025

Operating status

Operating

Headcount range

11–50 employees

Short description

BottleCap AI is a Prague-based startup developing efficiency-first foundational LLMs with proprietary non-transformer architectures. Founded by word2vec creator Tomas Mikolov and Beat Saber co-creator Jaroslav Beck, it ships its CAP1 model through consumer apps like Pulse: Community News and plans enterprise subscriptions.

Ownership category

akta.pro rank

Product category

Foundational AI Models

akta.pro industry

PRIMARYHDAEANAD

LLMOps & Generative AI Platforms (Prompt/Agent Orchestration, RAG)

Hardware, Technology & Telecom/Enterprise Platforms & Databases/Artificial Intelligence Platforms for Enterprise

SECONDARYHDAAACAD

Prompt Engineering, Orchestration & LLMOps Tooling

Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/Foundation Models & Generative AI (LLMs, multimodal)

SECONDARYHDAEAGAA

Document Management Systems (DMS)

Hardware, Technology & Telecom/Enterprise Platforms & Databases/Enterprise Content Management (ECM) & Document Management

Keywords

• Large language models
• Foundational AI models
• News aggregation app
• Efficient AI models
• AI research

Headquarters

HQ city

Prague

HQ country

Czechia

HQ region

Europe

Offices2 records

Markets served

GTM type

B2C

Offering type

Software

Cost components

Personnel, Technology or R&D, Operations, Marketing or Sales, Infrastructure

Revenue model

1. Consumer Applications: Revenue generated from consumer mobile applications distributed via app stores. The first product is Pulse: Community News on iOS. The company monetizes through in-house consumer apps rather than selling models directly.
2. Enterprise Applications: Planned future commercialization through enterprise applications built on their CAP1 foundational model. The company plans to commercialize through in-house applications rather than model licensing.

Go-to-market motion2 records

Distribution channels2 records

Marketing channels3 records

Core offering

BottleCap AI builds efficiency-first foundational large language models (LLMs) using novel architectures beyond transformers, with its CAP1 model targeting 100x efficiency improvements and 50% lower training costs. The company commercializes this technology through in-house consumer applications, starting with Pulse: Community News — an AI-powered news aggregation app that surfaces real-time community sentiment on current events. Future commercialization is planned through enterprise applications built on top of CAP1.

Product overview

BottleCap AI is developing a portfolio of efficiency-first large language models and applications. The core offering is the CAP1 foundational LLM, a proprietary architecture-first language model designed to achieve significant efficiency improvements over transformer-based models. The company commercializes through consumer applications, with the first product being Pulse: Community News—an AI-powered news aggregation iOS app that runs on CAP1 and provides both news aggregation and real-time community sentiment insights.

Differentiator

Problem solved

Functional benefit

Brands

• Pulse: Community News: AI-powered news aggregation iOS app built on the CAP1 foundational model, designed to help aggregate information and provide community insights on current events.

Products and services

• Pulse: Community News AI-powered mobile news aggregation application that summarizes news headlines and provides real-time community sentiment insights on how people are reacting to specific events. Built on top of the CAP1 foundational model and distributed via the iOS App Store and Google Play Store for individual consumer users who want to distinguish important signals from news noise.

Quantifiable outcome

• 50% reduction in LLM training costs
• +2 more outcomes

Segments2 records

Ideal customer profiles2 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

AI capability2 records

Feature3 records

Scale indicators3 records

Recent moves6 records

Expansion highlights5 records

Direct peers

• Aleph Alpha: Heidelberg-based European foundational model startup with sovereign/regulated focus; comparable in European positioning, enterprise/government targeting, and funding stage dynamics.
• AI21 Labs: Foundation model company building proprietary LLMs (Jamba) with novel architectures (including SSM hybrid models); shares the architectural-innovation angle and enterprise commercialization pathway.
• Mistral AI: Paris-based foundational LLM startup building efficient open and commercial models; the most directly comparable peer given similar European origin, foundational-model focus, and product-led commercialization strategy.
• DeepSeek: Chinese AI lab that achieved global attention by demonstrating extreme cost-efficiency in training frontier-grade models; closest philosophical match to BottleCap's architecture-first, efficiency-first thesis.
• Cohere: Enterprise-focused LLM company with proprietary foundational models and a focus on retrieval-augmented and efficient deployment; comparable on custom-architecture LLM + enterprise GTM.

Emerging players

• Inflection AI: Foundation model startup that pivoted from consumer Pi to enterprise; shares the consumer-then-enterprise commercialization trajectory and small-team research model.
• Hugging Face: Open AI platform and model hub, also a BottleCap investor; overlapping ecosystem role but primarily infrastructure/distribution rather than direct model competitor.

Broad incumbents

• OpenAI: Dominant foundational model provider and the explicit competitor referenced by BottleCap's founders; broader product portfolio but the benchmark BottleCap must beat on efficiency.
• xAI: Elon Musk's AI lab building Grok with massive compute; relevant incumbent benchmark for any efficiency-focused foundation model competitor.
• Anthropic: Frontier foundation model lab with Claude; broader portfolio and substantially more capital, but directly comparable in building efficient, safety-focused proprietary LLMs.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks5 records

Key highlights6 records

Customer concentration

• Xx.com/bottlecapai
• LinkedInlinkedin.com/company/bottlecap-ai

Revenue estimate

Valuation estimate

Number of profiles

Profiles3 records

Funding overview

Funding rounds1 record

Investors8 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about BottleCap AI