California-based tech company Gleen has secured fresh funding in an oversubscribed round, attracting investors from both the established software world and the nascent crypto sphere. Gleen, founded by Microsoft and LinkedIn veterans, offers an enterprise-grade chatbot aimed at technical communities on platforms like Discord and Slack. The company raised $4.9 million, with institutional investors such as Coinshares and angels like Solana co-founder Anatoly Yakovenko and former Chainlink COO Mike Derezin participating.
Key Takeaway
Gleen’s chatbot addresses the issue of hallucination, where artificial intelligence generates false but confident answers. By leveraging its proprietary machine learning layer and cross-checking responses with enterprise knowledge, Gleen’s chatbot provides more accurate and reliable answers. The company’s unique approach sets it apart from competitors that rely solely on large language models, making it an attractive solution for highly technical communities and companies.
Focusing on Hallucination: the Urgent Problem to Solve
Gleen’s narrow focus on solving the problem of hallucination in large language models sets it apart from other chatbot services. Hallucination occurs when AI generates false information but presents it with confidence, posing a risk in discussions surrounding esoteric topics. For instance, misinformation about token prices in a blockchain community could result in market manipulation. Thus, Gleen prioritized fixing this issue before establishing itself as a serious player in the market.
Attracting Diverse Investors
Gleen’s vision and innovative approach garnered interest from investors across the software and crypto landscapes. Institutional investors include Slow Ventures, 6th Man Ventures, South Park Commons, Spartan Group, and Coinshares. Notably, Solana co-founder Anatoly Yakovenko and former Chainlink COO Mike Derezin joined as angel investors in the funding round.
Gleen’s Unique Solution: Cross-Checking Responses with Enterprise Knowledge
To combat hallucination, Gleen developed its own proprietary machine learning layer that sources from enterprise knowledge. This approach involves cross-checking responses from large language models like OpenAI, Anthropic, or finetuned Llama, ensuring the most accurate answers based on domain knowledge. Gleen’s model has been trained on 100,000 pairs of questions and answers, enabling it to provide reliable responses.
Gleen’s Secret Sauce: Proprietary Search Algorithm
Gleen’s secret sauce lies in its proprietary search algorithm, fundamental to its system. This algorithm, combined with its ability to glean data from various sources such as knowledge bases, forums, and Slack or Discord discussions, allows Gleen to abstract information effectively without the need for excessively clean documentation. This unique approach makes Gleen particularly appealing to technical communities where accurate answers are crucial.
Expanding into the Enterprise Market
Initially popular among web3 customers, Gleen now generates more revenues from non-crypto users. With a team of eight employees, Gleen currently serves over 10 customers who pay based on the number of conversations generated for the chatbot. The company is gradually shifting its focus towards the medium-sized enterprise market, aiming to eventually solve customer service problems for a wide range of businesses, regardless of their industry.
Future Plans and Growth Strategy
Gleen intends to utilize its fresh funding to further develop its chatbot, drive sales, and execute marketing strategies. The company aims to educate users about issues related to hallucination, security, and compliance in the field of generative AI. While Gleen plans to have a sales force, it believes that inbound channels will remain the primary source of customer acquisition. As an AI-powered solution, Gleen acknowledges the ever-changing nature of technology and is committed to staying adaptable to ensure it continuously solves customer support problems in the most effective way possible.
