2024 October-November Analysis Featured

Enagram: Creating Georgian Digital Resources for the Gen AI Era

Investor.ge interviewed Rati Skhirtladze, CEO and Founder of Data Analysis Laboratory, about the team’s AI-powered platform, Enagram.ai, and the critical need to digitalize low-resource languages like Georgian.

As the world increasingly turns to AI and digital technology, low-resource langauges like Georgia are struggling to stay relevant across sectors and industries. The Georgian-created Enagram.ai,an AI-powered platform, is using creative methods to bolster the language’s online presence and provide 21st-century digital solutions for Georgian speakers.

The global AI landscape presents challenges, especially for languages like Georgian that lack digital resources.

“LLMs [large language models] like ChatGPT and Gemini, are based on large datasets in the English language. Creating an LLM for the Georgian language is very challenging because there simply aren’t enough digital resources available to train a language model,” Data Analysis Laboratory CEO and founder, Rati Skhirtladze, explains.

UNESCO estimates that at least 50 percent of the world’s more than six thousand languages are losing speakers and, in most world regions, about 90 percent of the languages may be replaced by dominant languages by the end of the 21st century. The Georgian language is also under a significant risk, although not entirely on the line to extinction, according to Skhirtladze. “The Georgian language is also under threat, although it is relatively protected due to its status as the official state language,” he says. “However, if Georgian is not equipped with modern technologies, it too will face risks.”

The challenges facing languages like Georgian in the digital world are significant. While LLMs like ChatGPT work well with high-resource languages like English and Spanish, languages like Georgian lack such digital material, creating a gap that threatens its future competitiveness in the global technological landscape.

“This is evident in the fact that when leading researchers at Google publish a paper, speakers of high-resource languages, like English or Spanish, can immediately use and apply it, resulting in new publications and products,” Skhirtladze says. “However, when you don’t have these resources for low-resource languages, you can’t create similar products.”

Developing an LLM of that scale and quality is extremely difficult. Nevertheless, Enagram’s team has approached these limitations with creativity—crafting custom datasets for their language tools, even hiring voice actors to build their text-to-speech functionalities. It’s also important to note that the architecture of LLMs allows for high-accuracy results in specific domains, such as the medical or banking fields. These principles and architecture can achieve excellent results for narrow tasks, which is what the DAL Team is working on.

Enagram is the culmination of over 20 years of experience and research in language technologies, led by a team of dedicated experts. The platform’s founders began their work with traditional grammar systems. Before Enagram, in 2006 Rati Skhirtladze and a friend started Ena.ge, an extensive digital platform offering language resources for the Georgian language. That project slowly transitioned to AI-powered solutions as artificial intelligence evolved. They developed the first Georgian spell-checker integrated into Microsoft Word, building upon their experience to craft innovative tools like the platform’s highly accurate text-to-speech tool.

CEO Rati Skhirtladze (PHOTO BY DataFest TBILISI 2024)

Skhirtladze says the platform continues to grow, adding new modules and capabilities regularly. Looking forward, Enagram plans to enter the U.S. market, as well.

The AI technologies on the platform have wide-ranging applications across industries. In the medical field, for instance, doctors can dictate notes and documents through speech recognition, saving time and enabling them to see more patients. Globally, professionals are turning to AI tools to maintain a competitive edge, and Enagram offers Georgian users access to these groundbreaking advancements, helping businesses and individuals alike work smarter, not harder.

For example, text-to-speech capabilities can help people with impaired vision access information that would otherwise be unavailable, broadening their engagement with the world.

Additionally, tools like speech-to-text allow students to easily transcribe lectures, offering them the flexibility to review and search specific content quickly and efficiently.

Enagram’s significance lies in its potential to make things easier at work and in private life, Skhirtaladze says.

“This is especially true for people with disabilities, particularly for the blind…for whom sound is the only source of receiving information, as Braille is quite limited,” he explains. “Social media can use this text-to-speech technology, which was impossible before the development of artificial intelligence.”