Box announces unlimited queries for Box AI and coming integration with GPT-4o – Business

Box Inc. today is enhancing its popular AI service that helps to increase the productivity of business workers, adding new features that will make it more powerful.

The cloud content management company also announced unlimited queries for Enterprise Plus subscribers in order to meet the growing demand for Box AI’s capabilities. Box AI is a suite of generative artificial intelligence tools that help business workers to ask questions about their data stored in Box and create new content on the fly, such as marketing emails or newsletters.

In an interview with SiliconANGLE, Box Chief Executive Aaron Levie said AI will transform everything about how people work in the future, and content is at the center of that. He added that the greatest value will be derived from unstructured information, which accounts for about 90% of all enterprise data. “That means we have all of this untapped value,” he explained.

Box AI is helping companies to tap into that value already in a myriad of ways, Levie said. The company launched Box AI last year, and since then the service has been adopted by hundreds of enterprise customers that want to tap the capabilities of generative AI to make their workforces more productive.

For instance, life sciences companies are using Box AI to extract insights from clinical studies to accelerate drug development, while financial firms use it to analyze documents to try and uncover trends on things such as stock prices. It’s also popular with marketing teams, which use the platform to quickly generate blogs and social media posts.

“We saw a lot more demand than even we expected,” Levie said. “It’s the core conversation I’m having with every customer. Everybody is looking for practical ways to use gen AI.”

Unlimited queries

When Box AI first launched, Enterprise Plus subscribers were limited to a rather paltry 20 queries per user per month, but demand for its capabilities has far exceeded that, Levie said. The company is now rectifying that, enabling unlimited queries for premium customers.

“Token prices have gotten cheaper in the last few months, so that enabled the unlimited feature,” Levie said.

International Data Corp. analyst Amy Machado said the unlimited queries will likely convince more customers to adopt Box AI. She explained that some enterprises are hesitant to invest in AI company-wide or at the individual user level because of concerns around costs and other factors.

“By removing these query caps for Box AI, Box can lower the barrier of entry to AI, making it easier for businesses to integrate and benefit from this powerful technology,” Machado said.

The unlimited queries for Enterprise Plus subscribers are available now, and users will have much more to look forward to in the coming months, for the company has a number of updates in the pipeline.

Feature enhancements

The most promising new feature is Box AI’s coming integration with OpenAI’s most powerful large language model, GPT-4o, which was launched last month. GPT-4o adds multimodal capabilities, which means it – and therefore Box AI – will be able to deliver better answers, with support for 20 additional languages that previously weren’t available.

Box AI will also gain support additional file types beyond what it can currently deal with. For instance, it will be able to handle natural language queries for image files. With this, users will be able to ask specific questions about any image stored in Box, or query information on spreadsheets.

Last, there’s an all-new metadata application programming interface that developers will be able to use to integrate Box AI with custom applications, available in beta for Enterprise Plus users now. Those apps will then be able to extract information from documents stored within Box, such as key data fields from unstructured documents, and automate various processes by leveraging file metadata.

7

Levie this will enable third-party applications to leverage the capabilities of Box AI. He gave the example of a logistics company that deals with bills of lading that contain all of the shipping information for a particular order. Box AI can be integrated with that logistics firms’ system to quickly find information related to a specific shipment or shipping company.

“AI is good at reading documents and extracting information,” Levie said. “Tell it to please extract the shipper’s address, and it does that instantly.”

Paving the way for agentic AI

As Box AI becomes more intelligent and further streamlines integration with third-party apps, it’s bringing the company closer to its goal of creating an “agentic AI” system that can act independently and proactively to complete more complex tasks for business users.

Agentic AI refers to the idea of futuristic AI systems that can set their own goals, learn from interactions and make decisions on their own. For instance, while a traditional generative AI chatbot can help to create a script for customer support and generate personalized responses to callers based on their inputs, agentic AI promises to go further than this.

Eventually, the goal is that such systems will be able to handle customer queries in a fully autonomous way, resolving complicated issues from start to finish without any human input. Agentic AI systems are those that can mimic the reasoning, execution and course-correction mechanisms used by humans.

“We’re very early. It’s day one of agentic AI, especially in an enterprise context,” Levie said. “But we’re seeing early signs that this is very real. Once you have agents that can talk to each other to do a task, you begin to be able to solve a broader set of enterprise tasks.”

In other words, the vision of agentic AI has not yet been realized. One of the main challenges is that the AI models that will power this are not yet capable of fulfilling every kind of work-related task. The other problem is making agentic AI cost-effective.

Fortunately, Levie said he sees a viable path forward, with aspects such as Moore’s Law, advances in software efficiency and more specialized AI models potentially reducing the costs of agentic AI by 10 times or even 100 times in the future.

“Agentic AI will be disruptive for companies that don’t take advantage of AI yet,” Levie said. “You can imagine telling an AI to do much more than simple, interactive work. You’ll be able to ask it to do multiple things in a workflow.”

With reporting by Robert Hof

Image: Box