The latest State of AI in the Enterprise report from Box is out! One year ago, 8% of organizations described themselves as advanced or leading edge in AI. Today that figure is 64%. But adoption is not the same as impact. Half of leading edge companies report significant ROI. Among early stage organizations it is just 1 in 9. We surveyed 1,640 IT decision makers across the US, UK, France, and Japan to find out.

The Box MCP connector is now available in @Harvey's Connector Library to early access customers. Search files, query documents, extract metadata, and create content in Box directly from Harvey, with existing permissions and security controls enforced throughout. Watch it in action.👇

What should security leaders actually take from the Fable 5 release? Our CISO Heather Ceylan breaks it down. How the safeguards work in practice. What the 30-day data retention condition means for your governance. And why the programs that win will be the ones already built to act on what these models find, not the ones that get access first. Read here. 👇 blog.box.com/what-security-l…

What does it look like when a technical advisory firm uses Box AI to rethink how it works with clients? CNM automated their entire MSA process using Box Relay, Box Metadata, and Box Doc Gen, giving every internal team real-time visibility into contract status from generation to signature. Watch here and learn more at box.com/ai 👇

This is exactly what we mean when we talk about the company brain. @chriskim_dev gave @NousResearch Hermes agent access to Box as its single source of truth. Check it out.👇

Most AI security efforts focus on what an agent is allowed to access. Our CISO Heather Ceylan argues that is no longer enough. As agents delegate to other agents and operate across multi-step workflows with less oversight, the real risk shifts from access control to execution control, how agents behave once they are acting, not just what they can touch. Read here. 👇 blog.box.com/rethinking-trus…

Every agent framework wants to be the next React for agents. But what if the best interface for agent memory already exists, and every LLM has already been trained on it to the tune of trillions of tokens? @crtr0 makes the case for file systems as agent memory. 👇

Manual invoice processing is slow, error-prone, and expensive to scale. Every PDF that lands in an inbox needs someone to open it, read line items, and key data into a spreadsheet or ERP. Box Extract pulls structured fields like vendor name, invoice number, totals, and dates directly from a PDF, and writes them back as searchable metadata, ready for ERP integration. Learn more here. developer.box.com/guides/tut…

Unfortunately, Claude Fable will no longer be available in the Box AI Studio. We hope this issue gets resolved and we can offer it again to customers soon.

Most enterprise AI programs fail not because the models are weak. They fail because organizations invest in AI before they have the data, governance, content, and execution readiness to support it. Our guide breaks down what a real enterprise AI strategy looks like, why most programs stall at the pilot stage, and the seven steps to build one that actually delivers at scale. Read here.👇 blog.box.com/enterprise-ai-s…

.@Box Agent is built on Deep Agents. ✅ Cross-library search ✅ Multi-doc synthesis ✅ Structured reports ✅ All within Box's existing security and permissions model

When Box transitioned Box AI into a multi-step agentic system, we chose @LangChain's LangGraph over building a custom execution engine from scratch. Our full architectural breakdown covers how agents are defined and compiled at runtime, how the Global Agent spawns sub-agents dynamically, how LangGraph's checkpointing powers session resumability, and how Box enforces user-scoped security at the platform level, not the LLM level. Read here.👇 blog.box.com/how-box-built-i…

At Box, we just surveyed 1,640 IT leaders across the US, Japan, and Europe about agentic AI adoption. Many standout findings, but a big one was that the companies that adopted AI the most are planning to grow headcount the most. Obviously lots of ways you can read that data and variables mixed in, but it’s actually quite intuitive that the companies that become most productive want to (and are able to) reinvest back into the business to keep getting the gains going. The narrative of jobs being wiped out assumes that companies will take a fixed approach to what they want to be able for work on. What’s happening in practice is it’s causing companies to want to light up more engineering projects, sell to more customers, automate more processes to give time back, and more. That all leads to more work to be done by people.