It’s clear that there is a shift taking place. Companies need to think about how they can put large language models into action. But, as with any advanced technology, this is easier said than done.
A startup in its early stages is at the perfect place at the perfect time to help companies leverage these new capabilities and build AI-enabled software on top large language models. The company announced today that it closed a $7 million seed funding round at the start of last year. Alex Halliday, the CEO and co-founder of the company says that there is a big challenge for businesses to get involved with LLMs. “There is still a big gap between the amazing features that people can use in ChatGPT compared to what businesses can do.” [applying that] The kind of challenges that are the hardest in the business. So we’re creating an open platform that allows people to create custom solutions over these algorithms that can really move numbers within the business,” Halliday explained to TechCrunch. The company currently helps customers build applications based on three LLMs, GPT-4 GPT-3 and Claude. The company says that the idea is to enable users to automate processes, extract data insights, generate personalized content and perform natural language techniques. Halliday states that customers are interested in leveraging their data and content with LLMs so that they can create new content based on their existing corpus, or create a generative AI-based experience on top their existing software. The company has a primary value proposition that helps customers use the models more efficiently because it can be costly. “What’s really interesting is that the larger models can be used to train smaller models. For the first few months, perhaps, you would use GPT-4 to train a larger, open-source model that has been finely tuned,” he said. AirOps is a tool that can help you navigate through these steps. “We’re learning the right architectures and recipes here, but over time, we expect the kind of boil the ocean, sledgehammer, approach will give way a more nuanced understanding of how people can take advantage of the options they have,” he said. The company was founded last year to help organizations get the most out of their data. But as LLMs became more popular, it shifted its focus. “As we began to examine the application of LLMs in the data space, it became clear that there was a much greater opportunity to help people blend LLMs and their data to create customized workflows and applications,” said he. Last fall, they shifted their focus towards this approach. The company has 14 staff with a few positions still open. Halliday said he sees diversity in many dimensions. He is also aiming to build an employee base that is diverse as he builds his company. This is especially true considering how new LLMs tend to be. He said that they were “very open-minded” when it came to hiring people from different backgrounds. Wing VC led the $7 million seed investment with participation from Founder Collective XFund Village Global Apollo Projects Lachy Groom.