Tag: cta__logo

On Thursday, Box launched its developer conference BoxWorks by announcing a new set of AI features, building agentic AI models into the backbone of the company’s products.

It’s more product announcements than usual for the conference, reflecting the increasingly fast pace of AI development at the company: Box launched its AI studio last year, followed by a new set of data-extraction agents in February, and others for search and deep research in May.

Now the company is rolling out a new system called Box Automate that works as a kind of operating system for AI agents, breaking workflows into different segments that can be augmented with AI as necessary.

I spoke with CEO Aaron Levie about the company’s approach to AI, and the perilous work of competing with foundation model companies. Unsurprisingly, he was very bullish about the possibilities for AI agents in the modern workplace, but he was also clear-eyed about the limitations of current models and how to manage those limitations with existing technology.

This interview has been edited for length and clarity.

You’re announcing a bunch of AI products today, so I want to start by asking about the big-picture vision. Why build AI agents into a cloud content-management service?

So the thing that we think about all day long — and what our focus is at Box — is how much work is changing due to AI. And the vast majority of the impact right now is on workflows involving unstructured data. We’ve already been able to automate anything that deals with structured data that goes into a database. If you think about CRM systems, ERP systems, HR systems, we’ve already had years of automation in that space. But where we’ve never had automation is anything that touches unstructured data. 

Techcrunch event

Join 10k+ tech and VC leaders for growth and connections at Disrupt 2025

Netflix, Box, a16z, ElevenLabs, Wayve, Sequoia Capital, Elad Gil — just some of the 250+ heavy hitters leading 200+ sessions designed to deliver the insights that fuel startup growth and sharpen your edge. Don’t miss the 20th anniversary of TechCrunch, and a chance to learn from the top voices in tech. Grab your ticket before Sept 26 to save up to $668.

Join 10k+ tech and VC leaders for growth and connections at Disrupt 2025

Netflix, Box, a16z, ElevenLabs, Wayve, Sequoia Capital, Elad Gil — just some of the 250+ heavy hitters leading 200+ sessions designed to deliver the insights that fuel startup growth and sharpen your edge. Don’t miss the 20th anniversary of TechCrunch, and a chance to learn from the top voices in tech. Grab your ticket before Sept 26 to save up to $668.

San Francisco
|
October 27-29, 2025

REGISTER NOW

Think about any kind of legal review process, any kind of marketing asset management process, any kind of M&A deal review — all of those workflows deal with lots of unstructured data. People have to review that data, make updates to it, make decisions and so on. We’ve never been able to bring much automation to those workflows. We’ve been able to sort of describe them in software, but computers just haven’t been good enough at reading a document or looking at a marketing asset.

So for us, AI agents mean that, for the first time ever, we can actually tap into all of this unstructured data.

What about the risks of deploying agents in a business context? Some of your customers must be nervous about deploying something like this on sensitive data.

What we’ve been seeing from customers is they want to know that every single time they run that workflow, the agent is going to execute more or less the same way, at the same point in the workflow, and not have things kind of go off the rails. You don’t want to have an agent make some compounding mistake where, after they do the first couple 100 submissions, they start to kind of run wild.

It becomes really important to have the right demarcation points, where the agent starts and the other parts of the system end. For every workflow, there’s this question of what needs to have deterministic guardrails, and what can be fully agentic and non-deterministic. 

What you can do with Box Automate is decide how much work you want each individual agent to do before it hands off to a different agent. So you might have a submission agent that’s separate from the review agent, and so on. It’s allowing you to basically deploy AI agents at scale in any kind of workflow or business process in the organization.A Box Automate workflow, with AI agents deployed for specific tasks.Image Credits:Box

What kind of problems do you guard against by splitting up the workflow?

We’ve already seen some of the limitations even in the most advanced fully agentic systems like Claude Code. At some point in the task, the model runs out of context-window room to continue making good decisions. There’s no free lunch right now in AI. You can’t just have a long-running agent with unlimited context window go after any task in your business. So you have to break up the workflow and use sbagents.

I think we’re in the era of context within AI. What AI models and agents need is context, and the context that they need to work off is sitting inside your unstructured data. So our whole system is really designed to figure out what context you can give the AI agent to ensure that they perform as effectively as possible.

There is a bigger debate in the industry about the benefits of big, powerful frontier models compared to models that are smaller and more reliable. Does this put you on the side of the smaller models?

I should probably clarify: Nothing about our system prevents the task from being arbitrarily long or complex. What we’re trying to do is create the right guardrails so that you get to decide how agentic you want that task to be.

We don’t have a particular philosophy as to where people should be on that continuum. We’re just trying to design a future-proof architecture. We’ve designed this in such a way where, as the models improve and as agentic capabilities improve, you will just get all of those benefits directly in our platform.

The other concern is data control. Because models are trained on so much data, there’s a real fear that sensitive data will get regurgitated or misused. How does that factor in?

It’s where a lot of AI deployments go wrong. People think, “Hey, this is easy. I’ll give an AI model access to all of my unstructured data, and it’ll answer questions for people.” And then it starts to give you answers on data that you don’t have access to or you shouldn’t have access to. You need a very powerful layer that handles access controls, data security, permissions, data governance, compliance, everything. 

So we’re benefiting from the couple decades that we’ve spent building up a system that basically handles that exact problem: How do you ensure only the right person has access to each piece of data in the enterprise? So when an agent answers a question, you know deterministically that it can’t draw on any data that that person shouldn’t have access to. That is just something fundamentally built into our system.

Earlier this week, Anthropic released a new feature for directly uploading files to Claude.ai. It’s a long way from the sort of file management that Box does, but you must be thinking about possible competition from the foundation model companies. How do you approach that strategically?

So if you think about what enterprises need when they deploy AI at scale, they need security, permissions, and control. They need the user interface, they need powerful APIs, they want their choice of AI models, because one day, one AI model powers some use case for them that is better than another, but then that might change, and they don’t want to be locked into one particular platform.

So what we’ve built is a system that lets you have effectively all of those capabilities. We’re doing the storage, the security, the permissions, the vector embedding, and we connect to every leading AI model that’s out there.

When Seth Winterroth left his job at GE Ventures to help launch Eclipse in 2015, robotics was on his mind. Or more specifically, the number of early-stage robotics startups that were struggling to launch due to lack of interest.

“These are teams that had just finished their postdocs at Waterloo, or CMU, or MIT and were starting robotics companies, and the refrain that I continually heard from the startups was, ‘Hey, we’re having a really hard time raising institutional venture capital,’” Winterroth told TechCrunch. “At the time in Silicon Valley, most venture capital was going into the very mature application layer or the application layer of some very mature computing platforms.”

A lot has changed since then.

Now, after investing in robotics startups for 10 years, Winterroth, a partner at Eclipse, said the time to invest in robotics has never been better. The robotics startup market has matured and the hardware and software powering these bots has gotten significantly better — and cheaper.

Venture investing in the category is gaining momentum as well. Investors poured $6 billion into robotics startups in the first seven months of 2025 according to Crunchbase data. The data company predicts that this year’s funding totals will eclipse 2024, making it one of the only non-AI categories to experience a boost in funding.

While one could argue that robotics is seeing a surge in investor interest because of AI — and it’s not wrong to acknowledge AI’s role in the advancement of robotic tech — investors who have focused on the category longer than the last few years said the industry didn’t get to this point just because of advancements in AI over the past few years.

Reaching maturation

The real catalyst for the industry to start gaining momentum actually happened back in 2012, Winterroth said, when Kiva Systems, a small startup based out of Massachusetts, got acquired by Amazon.

Techcrunch event

Join 10k+ tech and VC leaders for growth and connections at Disrupt 2025

Netflix, Box, a16z, ElevenLabs, Wayve, Sequoia Capital, Elad Gil — just some of the 250+ heavy hitters leading 200+ sessions designed to deliver the insights that fuel startup growth and sharpen your edge. Don’t miss the 20th anniversary of TechCrunch, and a chance to learn from the top voices in tech. Grab your ticket before Sept 26 to save up to $668.

Join 10k+ tech and VC leaders for growth and connections at Disrupt 2025

Netflix, Box, a16z, ElevenLabs, Wayve, Sequoia Capital, Elad Gil — just some of the 250+ heavy hitters leading 200+ sessions designed to deliver the insights that fuel startup growth and sharpen your edge. Don’t miss the 20th anniversary of TechCrunch, and a chance to learn from the top voices in tech. Grab your ticket before Sept 26 to save up to $668.

San Francisco
|
October 27-29, 2025

REGISTER NOW

“I like to say Kiva Systems’ acquisition was the acquisition that launched 1,000 robotic startups,” Winterroth said. “Between 2011 and 2015, 2016, that really was the case. You just saw a number of different new companies get started. Some like 6 River Systems, or Clearpath Robotics, were successful, but most were not. But that talent learns and that learning compounds, and it’s brought into the next set of ventures.”

This first wave helped attract engineers to the sector and helped companies figure out product-market-fit, he said.

Kira Noodleman, a partner at Bee Partners, echoed this. Noodleman told TechCrunch the last decade of trial and error helped startups figure out what the market is actually looking for when it comes to robotics and automation.

Some companies, like Rapid Robotics, which Noodleman backed, shut down trying to figure out what the market wanted. Those failures have helped the next batch of startup founders, who now have a much better idea of what potential customers want from this sector.

Noodleman had a similar experience with her own investing thesis, she said, which changed as the market matured.

“Lights out manufacturing assumes there are zero humans in the loop; that is just not happening. We proved that already back in the 2010s,” Noodleman said. “Let me take a simple task, machine tending, all it is is someone’s hand putting something in and out of a machine. The point here is you can imagine how many low-hanging fruit, repetitive tasks there are, like machine tending.”

Fady Saad, a general partner at early-stage robotics-focused Cybernetix Ventures, also launched his firm prior to the AI boom after he noticed he was spending a lot of time connecting early-stage robotics companies to sources of funding during his time as a co-founder at MassRobotics.

Falling hardware costs have also driven investor interest in the sector, Saad said, noting that it’s cheaper to build robots today than five years ago. This allows companies to have a more viable path to scale and makes them more attractive to potential venture backers.

“The cost of building robotics has been going dramatically down,” Saad said. “Advances in sensor technology, compute, and batteries, all of that, it was the perfect timing to start full-stack robotics solutions.”

Advancements in AI aren’t hurting the industry either. While AI is being touted by many as the main reason why robotics are starting to see an increase in interest — alongside an Elon Musk-driven fascination with humanoid robots — it isn’t the only factor.

Saad added that while AI and large language models can be helpful for training robots, these LLMs are primarily trained on online information, whereas robots interact with the real world.

There are companies building models based on that real-life data; Nvidia just released a new set of world models for robot training in August. But Saad predicted it will take a bit longer to capture and train robots, especially those that will exist alongside people, on world data.

Present day

Momentum in the industry may be starting to swell, but that doesn’t mean every startup has figured out the best approach yet. Nor are some categories within robotics as mature as others.

Some of the first few markets to adopt robotics and automation, including manufacturing, warehousing, and construction, continue to be attractive for robotics startup backers.

For Winterroth, Saad, and Noodleman, healthcare and surgical-related robots remain a compelling area to invest in too. Noodleman adds eldercare to that category as well.

“In-home assistance is interesting, coming from me having looked at industrial robotics for 10 years,” Noodleman said. “Manufacturing and mining, burning labor shortages, aging populations, no humans are available at any price, even imperfect robotics are better than nothing.”

Saad added that vertically focused robotics companies tend to have access to more real-world and physical data, too, than horizontal players.

One area that these VCs are not as excited about are humanoids or consumer — and especially not consumer-focused humanoids.

Saad isn’t convinced that people will want to have a robot in their house anytime soon. He added that even non-humanoid consumer-focused robotics companies have struggled to get consumers excited.

“The only successful consumer robot company, iRobot, failed to come up with a second act,” Saad said. “Pool cleaning robot, lawn mower, mopping and floor-cleaning robot, none of these worked out for whatever reason.”

While the industry is still years away from commercial success of more intricate robotic models, like humanoids, VCs are pouring more capital into the sector. Despite the fact that this interest is driving up the costs of deals, the surge in interest is a net positive for the industry, Winterroth and Saad said, as the potential customer base for robotics startups continues to grow.

“There are enough examples of successful commercial organizations, successful robotics companies, that have become a valuable commercial organization,” Winterroth said. “Ten, 15 years ago, it was questionable whether or not there was going to be a large and thriving marketplace for these types of solutions. Now there’s a lot of customer awareness.”