Box

Banks have embraced AI. Their new challenge is proving how they used it. FINRA's 2026 Regulatory Oversight Report is clear: forensic logs, decision lineage, and human oversight are no longer optional. 3 steps to get there: -Prove every AI decision with forensic logs and lineage. -Rationalize your stack and stop fragmenting experiments. -Keep humans in the loop to validate every output. Full breakdown here. 👇 blog.box.com/ai-enters-fina…

.@ManusAI and Box CLI working together on a real workflow. Here's what that looks like. Your vendor documents like SOC2 reports and DPAs serve as the source of truth. HubSpot pulls in deal context. Manus analyzes everything, identifies risks and gaps, and saves a structured summary with recommended next steps right alongside the original files. Box is the end to end secure AI file system improving user productivity for real-life workflows. Watch here. 👇

Box Extract was built for the problem every document processing pipeline hits in production. Not a single API call. Not a prototype. A compliant, secure, end-to-end extraction pipeline with agentic workflows built in. See how it works. 👇 blog.box.com/beyond-just-wr…

Most enterprises are trying to drop AI agents into existing workflows and wondering why they're not getting the results they expected. @levie and @googlecloud's Rao Surapaneni break down how to rethink workflows for an AI-native world, how to govern agents safely across sensitive data, and what it actually takes to orchestrate multiple agents at enterprise scale. "I look at every employee as a manager of agents." Read the full recap here. 👇 blog.box.com/aaron-levie-an…

.@OpenAI just launched the new Codex with new features including computer use and image generation. Image creation is powered by gpt-image-1.5, and Codex can now do even more to help you build out and automate your enterprise workflows. Connect using the Box plugin and you can automate multi-step workflows and connect your tools together in ways that weren't possible before.

The new Codex is another jump in what agents will look like for knowledge workers. Agents that can code, work with tools, and use computers, can begin to execute long running tasks in the background for all areas of work. This can mean drafting reports, setting up data rooms for a merger, reviewing contracts, helping onboard clients, generating marketing assets, processing invoices, and more. With the Box plugin inside of the new Codex, you can begin to automate almost any kind of work with enterprise content. And importantly, being able to work across multiple apps is a huge point of leverage because we can now far more easily connect our tools together.

Watch the full episode here. 👇 youtube.com/watch?v=BEdIwT…

Most enterprise search is broken. The fix? A file system built for AI agents. Our CTO @BenAtBox breaks down why agentic search and context retrieval are transforming how enterprises find and use unstructured data in our latest AI Explainer episode.

Financial services firms have digitized their documents. Most still can't access what's inside them. Barnett Capital cut appraisal review from 1 hour to 1 minute. Valmark went from 4,000 extracted data points to 240,000. @USAA now mines complex vendor contracts for answers in seconds. Mercer Advisors reduced client onboarding from 2 weeks to 5 minutes. Same documents. Completely different outcomes. Full story here. 👇 blog.box.com/finserv-ai-doc…

97 sandbox agents. A company-wide AI certification program. A network of AI managers across every subfunction. Our COO Olivia Nottebohm shares the unfiltered story of what becoming AI-first actually looks like. "AI transformation is still human-led change." Read here. 👇 blog.box.com/coo-ai-transfo…

Box put @Anthropic's Claude Opus 4.7 to the test. Results: 56% fewer model calls, 50% fewer tool calls, 25% faster, 30% fewer tokens — with no loss in output quality. Since Box Agent consumes AI Units based on tokens used, that's less spend and more work. Claude Opus 4.7 will be available today in Box AI Studio and through the Box API. Details → blog.box.com/claude-opus-47…

Building reliable, production-ready document workflows just got a lot simpler. With the @OpenAI Agents SDK 2.0 and Box, developers get native sandbox support, enterprise-grade governance, and zero custom infrastructure to manage.

Want to connect AI agents like ChatGPT and Claude to your enterprise content in Box? We're breaking it all down in a live session and you should be there. The Box Extract API, the Box MCP server, example code, and best practices you can apply immediately. Register here. 👇 events.box.com/box-webinars/u…

Box developers can now build with the @OpenAI Agents SDK 2.0 — redefining what enterprise document workflows can do. AI assistants answer questions. AI agents get things done. Using the SDK's native sandbox support, developers can build autonomous agents that search, extract, and route documents inside Box — with zero-trust governance and no extra security setup. What once took days happens automatically, with a full audit trail. Box serves as the content layer in the Agents SDK stack, so compliance is built in from the start. Start building today with our open-source reference implementation on GitHub. 👉 bit.ly/4tkbcsY

Asking an AI agent for a file and getting a wall of text back is not collaboration. The Box MCP Server now brings rich file previews, multi-step workflows, and enterprise-grade security to ChatGPT and @glean Assistant — joining existing @claudeai support, with @Microsoft365 Copilot coming soon — so your teams can work smarter in the AI tools they already love. See what's new. 👇 bit.ly/4eo6vtf

Contract review shouldn't take days. Box Extract surfaces risk indicators instantly, and a custom AI agent suggests playbook-aligned redlines and maintains a complete audit trail, all within seconds. What took hours now happens instantly. Watch the demo. 👇 youtube.com/watch?v=FzlMpK…

Watch the full episode 👇 youtube.com/watch?v=giDpAw…

Most enterprises think non-deterministic AI outputs mean they can't trust agent workflows. Andrew Maas, VP of AI at @micro1_ai, disagrees and explains exactly how to engineer reliability into agentic systems on the latest Partner Podcast with our CTO @BenAtBox. Timestamps 02:54 What micro1 does and the role of human experts in AI systems 04:13 Rise of multi-step agentic workflows and domain-specific AI capabilities 07:48 Limits of current models and the need for deeper domain expertise 08:12 One-shot vs multi-step AI reasoning and why it matters 10:07 Composing multiple LLM steps to create reliable enterprise workflows 13:22 Variability in LLM outputs and concerns about enterprise reliability 18:54 Files as the new interface between humans and AI agents 22:24 Using evals and human review to improve AI systems in production 26:30 Experiment and challenge assumptions about AI limits

Building event-driven workflows that connect Box to your systems of record shouldn't require custom infrastructure. With Claude Managed Agents, it doesn't. Watch how an invoice upload triggers an agent that extracts vendor data and stores it directly in @databricks. See it in action. 👇

94% of organizations use AI, but only 37% are seeing real productivity gains. What separates them? In this deep dive, Box CTO @BenAtBox and @GoogleCloud's Rao Surapeneni break down exactly what's driving that gap. Here's what they cover: → How frontier AI translates to measurable business impact → What architecture successfully takes AI pilots to production → How leading enterprises in regulated industries are building governance frameworks that scale Watch now → events.box.com/box-webinars/a…