Contextual AI has come out of stealth, raising $20 million in seed funding for its pursuit of developing the next wave of language models for enterprise applications. Co-founded by Douwe Kiela, the technology aims to resolve issues limiting the adoption of large language models (LLMs) like OpenAI’s GPT-4 by enterprises with strict compliance and governance requirements.
Despite demonstrating the potential to revolutionize industries, existing LLMs are afflicted with limitations such as producing fabricated results with high confidence and difficulties in modifying their knowledge base. As a result, enterprises may hesitate to implement LLMs in their operations. To tackle these barriers, Contextual AI intends to create a new generation of LLMs that cater to specific enterprise needs.
Backed by investors like Bain Capital Ventures, Lightspeed, Greycroft, and SV Angel, Contextual AI's goal is to develop AI solutions that are more suitable for companies than consumer-focused LLM offerings. To that end, Kiela and his co-founder, Amanpreet Singh, rely on their background in AI development at Hugging Face and Meta.
Drawing upon the research experience of Kiela, who led efforts on retrieval augmented generation (RAG) while at Meta, the technology aims to develop a superior text-generating AI for enterprise uses. In essence, RAG works by enhancing LLMs with external sources, such as files and webpages, to improve their performance.
When provided with a question, RAG essentially searches for relevant data within these external sources, packaging the information with the original prompt and feeding it to an LLM. In turn, this generates a context-aware response, offering improved precision and reliability compared to conventional LLMs.
In addressing issues surrounding customization and attribution in LLMs, RAG promises improved performance while sidestepping the need for retraining or fine-tuning. Smaller, more efficient language models can now be developed, driving down latency and costs while delivering an integrated solution for enterprise requirements.
Although Contextual AI is not alone in this pursuit, as other startups like LlamaIndex also explore data injections and plug-in frameworks for LLMs, the company claims to have a unique advantage in the enterprise sector. Although currently pre-revenue, Contextual AI is reportedly in discussions with Fortune 500 companies for potential pilot projects of its technology.
By ensuring that the outputs of generative AI are accurate, reliable, and traceable, Contextual AI has the potential to make a significant impact on the adoption and deployment of AI technology in businesses. Moreover, this more integrated approach offers a competitive alternative to other generative AI companies looking to target the enterprise market.
In the no-code industry, platforms like AppMaster are driving growth by enabling businesses to build backend, web, and mobile applications with ease. Embracing solutions like no-code platforms can help reduce costs and accelerate development, especially as generative AI gains traction in the enterprise sector.
