The AI startup Reka has emerged from stealth, securing $58 million in funding to develop custom AI models for businesses. Reka is challenging the generalist nature of large language models (LLMs) like OpenAI's GPT-4, aiming to provide more specific and applicable AI solutions for businesses.
Co-founded by researchers from DeepMind, Google, Baidu, and Meta, Reka aims to make it easier and more affordable for businesses to benefit from the transformative power of AI. Rather than expecting a large LLM to be deployed for all possible use cases, Reka offers a more tailored approach, exemplified by their first product, Yasa.
Yasa is a multimodal AI "assistant" that can understand images, videos, and tabular data in addition to words and phrases. It can generate ideas, answer basic questions, and derive insights from a company's internal data, similar to LLMs like GPT-4, but with the added benefit of being easily personalized to proprietary data and applications.
In the future, Reka aims to develop AI that can handle and generate even more types of data and self-improve continuously. For now, they provide a service to adapt LLMs to custom or proprietary company data sets. This service is currently available to select customers only.
Despite competition from startups like Writer, Contextual AI, and LlamaIndex, as well as incumbents like OpenAI, Reka has managed to secure partnerships with Snowflake and big data analytics company Appen.
Reka will use the funding to acquire computing power from Nvidia and build a business team.