Box

The Agentic AI Era Is Here. So Is the Security Problem. Heather Ceylan, CISO at Box, is speaking at VB Transform 2026, where enterprise AI leaders tackle what autonomous systems actually demand from security. July 14–15, Hotel Nia, Menlo Park. Register here → https://bit.ly/497VGZt #VBTransform #AgenticAI #AISecurity

Claims systems of record are faster and more sophisticated than ever. But manual document handling still accounts for 60 to 70% of the total claim cycle and costs large carriers more than $100 million a year, according to Box analysis of carrier benchmarks. The problem is not the claims system. It is what happens before the record gets updated. Every loss run, medical record, police report, and attorney demand packet arrives as an unstructured document that someone has to open, read, and rekey into the system before anything can move forward. Box operates as the document intelligence layer between unstructured content and Guidewire Software ClaimCenter, classifying documents on arrival, extracting structured data, and routing exceptions for human review. Shelter Insurance expects payout times to fall from two weeks to hours by applying that approach across their claims operation. Read the full breakdown here. 👇 https://lnkd.in/gfQqy33Y

Our CISO Heather Ceylan has been watching the exploit window shrink for years. What changed recently is the slope. In late December 2025, a single attacker used AI models to identify vulnerabilities, generate exploits, and execute attacks across nine government agencies at a pace no human red team could match. Twenty unpatched CVEs. No zero-days. Just machine speed applied to an open patch window. According to Mandiant's M-Trends 2026 report, time-to-exploit has effectively gone negative. The CVE program published 48,185 new vulnerabilities in 2025. Defenders are triaging 130 disclosures a day against a response window measured in hours. The traditional patch cycle does not fit inside this curve and never will again. Heather's latest post lays out the three plays her team is running to operate at machine speed. Compressing exposure before a CVE lands. Remediating in a pipeline where AI authors candidate fixes and humans review rather than author. And detecting on post-exploit behavior rather than signatures. The head start defenders have is real. But it is shorter than it looks. How is your organization restructuring around this curve? Read the full analysis here. 👇 https://lnkd.in/g4E48zUr

We do not just build Box Automate. We use it. Box piloted Box Automate internally across three of our most document-intensive teams: legal, HR, and procurement. Here is what that looks like in practice: → Legal: contracts automatically transformed into structured, searchable data, replacing a $300,000 CLM tool → HR: onboarding and offboarding workflows running end to end with AI agents, no manual hand-offs → Procurement: vendor risk assessments now 60% faster, with report generation dropping from over an hour to five minutes The collective result is faster, more consistent work that is easier to govern at scale. Here is a look at how we are using Box Automate to build a truly AI-first Box. https://lnkd.in/gD4kVrZj

Law firms should be getting more out of AI than almost any other sector. The content is all there. Decades of filings, memos, contracts, and firm knowledge. Most are not getting that value. And the reason has nothing to do with AI technology. Adding AI to a broken content foundation does not fix the inefficiency. It moves it. When content is scattered across email threads, shared drives, and disconnected systems, AI tools find the wrong things, miss the right things, and operate outside the firm's control. Maninder Sagoo, VP Commercial Legal Team at Box and former private practice lawyer, writes about why law firms cannot scale AI without solving the content layer first. Including a real example from Fisher Phillips, where a manual OCG process that once relied entirely on email and spreadsheets is now automated end to end. Read the full breakdown here. 👇 https://lnkd.in/gtBQTwtH

Cisco's Collaboration Customer Success team manages a billion-dollar book of business across international agencies, with dozens of contractors spanning three global regions. Delivering consistent, high-quality customer success at that scale requires more than just good tools. It requires a seamless, secure, and intelligent ecosystem. By combining Box and Webex, Cisco transformed how their global team manages knowledge, drives adoption, and uses AI to work smarter across regions. How they achieved it: -A secure single source of truth: Box centralized thousands of files across three global regions, giving every contractor a consistent, standardized foundation to work from regardless of location. -Institutional knowledge made instantly queryable: With Box AI, the team can now query years of stored content in seconds, including a single PowerPoint with more than 150 revisions, moving from subjective opinions to data-driven insights instantly. -Automated workflows connected to their CRM: Using Box content and natural language prompting, the team built automation workflows that ingest Webex meeting summaries directly into CRM records and generate executive briefings automatically, eliminating hours of manual data entry. Frank Murphy, Director, Customer Success Management and Engineering, Collaboration, Cisco shared: "Instead of hunting through endless threads or relying on fuzzy memories to make decisions, our team could query content directly." For teams managing global operations at scale, Cisco's story is a compelling example of what becomes possible when content, AI, and workflows are connected in one secure environment. Read the full story here. 👇 https://lnkd.in/gmS-D4jc

Box is announcing support for Anthropic's latest model, Claude Opus 4.8, now integrated into Box AI to raise the bar for how enterprises manage and analyze content. We evaluated Claude Opus 4.8 head-to-head against Claude Opus 4.7 across rigorous, real-world enterprise use cases — and the results are clear. The wins are concentrated in structured, high-judgment work where missed details carry real consequences: → Report Drafting: Scored 87% (vs. 77%) on industrial goods report writing and 90% (vs. 84%) on consumer products launch evaluations. → Financial Services: Led by nearly 8 percentage points on a corporate lending analysis, extracting complex financial metrics with superior accuracy. → Public Sector: Achieved near-perfect accuracy on library grant data analysis and excelled in math intervention reporting. → Legal NDA Review: Identified more relevant clauses and flagged compliance issues with near-perfect consistency across independent runs. The pattern across these results points to something meaningful for enterprise AI: the performance delta compounds. In contracts, financial documents, and compliance workflows, reviewers often use model output as their starting point — which means gaps the model misses can become gaps no one catches. Opus 4.8's higher coverage and consistency directly reduces that risk. Claude Opus 4.8 will be available soon for Box AI customers in Box AI Studio and via the Box AI API. Read our full evaluation and get started: https://lnkd.in/g-Km7GiD

Most AI conversations start with the model. Our CTO Ben Kus thinks an adjustment is in order. In our latest AI Explainer, he breaks down why the first question every IT leader should be asking is not which AI model to use but where their data lives and whether the platforms holding it are accessible to agents in a headless way. AI agents interact with platforms differently than humans do. They aren’t opening web applications or mobile apps. They’re calling MCP servers, client APIs, and CLIs. And if your platform isn’t built to support those headless interactions, your agents won’t be able to access the data they need to do meaningful work. In the age of AI agents, the permissions, logging, storage, and compliance capabilities underneath your enterprise platforms are the foundation that determines whether your AI strategy actually delivers. Watch here. 👇 https://lnkd.in/gW724p9Z

The Forune Best Large Workplaces in the Bay Area list is here.🎉 To determine the ranking, Fortune partnered with global people analytics firm Great Place To Work to analyze survey feedback representing more than 70,000 U.S. employees. Top companies include: ➡️ Cisco ➡️ NVIDIA ➡️ Cadence ➡️ Salesforce ➡️ Box 🔗 Check out who made our full list: https://lnkd.in/effjWpUG

Workers' compensation claims processing is one of the most document-heavy workflows in insurance. Medical records, intake forms, supporting documentation, all of it reviewed manually before a claim can move forward. If your claims team is still working that way, there is a faster path. Join Box and Guidewire Software for a live demo and peer discussion on how to streamline claims processing workflows with AI-powered medical document extraction. You will see how Box and Guidewire can work together to automate workers' compensation claims handling end to end, hear what peers are seeing in the market right now, and connect directly with experts who can answer your specific questions. This is the kind of session where you leave with ideas you can actually bring back to your team. Register here. 👇 https://lnkd.in/gTgAvDF4