(Image credits: Techcrunch)
California-based startup Gleen is making waves in the world of chatbots with its unique approach to solving a pressing problem – hallucination. The company recently raised $4.9 million in an oversubscribed funding round, drawing support from both traditional software investors and crypto enthusiasts.
Founded by veterans from Microsoft and LinkedIn, Gleen is not your typical chatbot service. It has honed in on a niche market segment: technical communities. Specifically, Gleen is targeting communities like blockchain infrastructure channels on platforms such as Discord. This narrow focus has allowed the company to address a significant challenge faced by large language models (LLMs) in the chatbot industry.
Hallucination, in the context of artificial intelligence, occurs when a chatbot generates false information with unwavering confidence. In discussions involving complex and esoteric topics, such as cryptocurrency markets or blockchain technology, the consequences of hallucination can be dire. A single misleading but confidently delivered answer can sway market sentiment or mislead participants.
Ashu Dubey, co-founder and CEO of Gleen, explained the company’s dedication to solving the hallucination problem. “If someone says the price of Uniswap [the token of the namesake decentralized crypto exchange] is going to go up to $200, then it can be a massive manipulation of the market,” Dubey told TechCrunch in an interview. “We decided that we were going to fix this particular problem of hallucination before we could actually be a serious player, so we fixed it.”
Gleen’s approach to tackling hallucination sets it apart from many other chatbot services in the market. While some competitors simply wrap existing large language models, such as ChatGPT, Gleen has developed its proprietary machine learning layer. This layer draws from enterprise knowledge sources, enabling it to cross-check responses generated by LLMs to avoid hallucination.
Interestingly, Gleen’s reliance on LLMs accounts for less than 20% of its overall technology stack. The majority of the company’s work centers around how it stores and retrieves data using its proprietary system, ensuring that the responses are grounded in domain knowledge.
Gleen’s unique workflow involves cross-referencing responses from its proprietary machine learning layer with those from LLMs like OpenAI, Anthropic, or fine-tuned versions. This hybrid approach aims to deliver the most accurate and reliable answers to users. To train its model effectively, Gleen has utilized a dataset comprising 100,000 pairs of questions and answers.
Investors in Gleen’s recent funding round include Slow Ventures, 6th Man Ventures, South Park Commons, Spartan Group, and Coinshares, among others. Additionally, prominent figures in the blockchain and crypto space, including Anatoly Yakovenko, co-founder of Solana, and Mike Derezin, former COO of Chainlink, have lent their support as angel investors.
In a competitive landscape where chatbots often struggle with hallucination, Gleen’s innovative approach and commitment to accuracy position it as a promising player in serving technical communities and addressing the unique challenges they face in their discussions.
With its freshly acquired funding, Gleen aims to further refine its chatbot service and expand its reach within technical communities, ultimately contributing to more reliable and informed discussions in these specialized domains.
