We’ve seen a number of enterprise software companies making announcements related to generative AI in the last six weeks, from established companies like Salesforce and Adobe to startups like Ada and Forethought. Box is the latest to be announcing its own flavor of the technology under the moniker Box AI, and it will be teaming up with OpenAI to deliver the first tools on the platform.
Like many of the other companies making these announcements, they don’t have a product quite ready yet, but they are working with early customers to refine the functionality they have built in recent months. The company chose to call this Box AI because the two features being announced today are part of a much broader product road map of AI being added to the platform in the coming months.
Box CEO and co-founder Aaron Levie says that on one hand, the new AI features will help generate more content, but the real strength is helping people understand that content. “These large language models are uniquely good at reasoning through content. And so a lot of the use cases that we’re excited about are actually ones where you can use the large language model, not as the database of knowledge, but as a reasoning engine to work through your data,” Levie told TechCrunch.
For now the company is focusing on a couple of use cases in its initial announcement. For starters, you can click the Box AI button and ask questions of a document like, “Summarize this document for me,” or “What are five key points about this document?” And Box’s generative AI will provide an answer. This could come in handy for long reports or complex contracts.
Being able to iterate over a document and ask increasingly specific questions means that you can produce content like a quiz or summary and the AI should be able to improve as you ask more questions.
The other thing you can do is create new content in Box Notes. Say, you create 10 bullet points from a meeting about a new product. You could ask Box AI to make a blog post out of that list and it can do it pretty much instantly, working in a similar way to ChatGPT, OpenAI’s generalized generative AI tool.
The company is working closely with customers to refine these tools and Levie says that in time, you will be able to analyze multiple documents with refinements like the maximum age a document can be. Over time, you will also be able to build automated workflows, but these elements are on the product roadmap and are not part of this announcement.
Although, the company is working with OpenAI API for starters, the idea is to be flexible enough to accommodate any large language model, or even alternative model types that could develop over time.
Levie says he’s announcing these products now, even prior to beta, because he’s been getting a constant stream of questions from customers about the company’s plans for generative AI and he wants them to know what’s coming.