© Getty Images / iStockPhoto / MetaWorks / iStockPhoto.com
If companies understood the data they had better, he says, they could do more with it and save costs, too Martin Negsh. With its launch Field.ai He wants to help them. The startup has developed a solution that uses artificial intelligence to digitize documents, extract data, translate and summarize content, and automate processes.
The solution is already being used by a major German tax consultant, who uses it to handle their receivables management and fundraising for clients.
Use of artificial intelligence
To do this, thousands of messages must be recorded and evaluated every day. The system not only recognizes and maps data in tables or handwritten notes, but also summarizes the content if necessary or translates it if it is in another language. The solution is integrated into the usual working environment. “What previously had to be manually filled is now pre-filled,” says Negsh.
For this purpose, in addition to techniques, Text and object recognitionalso Language models used, such as those underlying the chatbot ChatGPT. Feld.ai uses the Big Language Model released by Facebook’s parent company, Meta Elma 2 It trains it itself using the company’s own data.
The data is processed and stored on a company-owned server in Vorarlberg FeldkirchWhere the company’s headquarters is located. The goal is to specifically target companies that do not want their data leaked to the outside world. “Everything stays in our data center,” Negsh says. Next year, Negesh hopes to be able to gain more customers for his solution.
➤ Read more: Mytalents.ai: Startups offer courses for dealing with artificial intelligence in companies
The entrepreneur, who previously worked for 15 years at a large Swiss reinsurer, founded the startup last year. It is financed in addition to the initial financing from the Development Bank austrian business service (aws) From sales revenue.
In addition to licensing income and volume components, the company also earns money from consulting services. “Because a lot of things have to be adjusted individually,” says Negsh. Any remaining money is put into hardware, such as Nvidia graphics cards, which are supposed to provide more performance when training AI models.
➤ Read more: Artificial Intelligence knows what you want to buy tomorrow
The founder says that finding investors is currently difficult. And also because generative AI was disappointed after the hype at the beginning of the year. “A lot of capital has been absorbed and now a lot of people have the idea that maybe there’s no way to make money from it.” Negsh says his startup is focused on improving processes: “We are not selling huge AI projects that promise a bright future. We are making processes more efficient.”
This article was created as part of a collaboration between Futurezone and austrian business service (aws).
“Total coffee aficionado. Travel buff. Music ninja. Bacon nerd. Beeraholic.”