March 2026 Summaries

Box's journey to becoming an AI-first company, as described by COO Olivia Nottebohm, highlights the complex process of integrating AI into business operations, emphasizing that transformation is not instantaneous but involves a series of strategic phases. These phases include ideation, universal education, scaling to production, and strategic redesign, each presenting challenges and learnings. The transformation relied heavily on fostering a culture of innovation, where employees were encouraged to experiment with AI in a secure, sandbox environment, leading to the creation of numerous AI agents to enhance productivity and unlock value from existing content. The initiative required comprehensive training and change management to ensure all employees engaged with AI, thus democratizing access and preventing a divide between early adopters and laggards. A critical insight was the necessity of curated, reliable content for agents to function effectively, prompting the establishment of structured governance and content management systems. As Box continued to refine its AI strategy, it became evident that human-AI collaboration offered significant efficiency gains, necessitating ongoing adjustments and a focus on training employees to adapt to new workflows enabled by AI.

Cleveland Construction, a family-owned commercial construction management firm, has significantly enhanced its operations by deploying 10 custom AI agents through Box AI Studio, transforming tasks like contract analysis and safety compliance and reducing contract review time from up to eight hours to just two minutes. The company's integration of AI into its workflows demonstrates the potential for construction firms to streamline processes and overcome document overload, which traditionally hampers productivity. By adopting intelligent content management and developing specialized agents such as a Contracts Agent, HR Agent, and Safety Agent, the firm has addressed specific construction workflows, leading to rapid efficiency improvements. Cleveland Construction's digitization efforts also include creating Box Hubs for easy file sharing and syncing with Autodesk, which has laid the foundation for further AI integration. Looking forward, the company plans an ERP migration in 2026 to enable comprehensive AI-powered analysis, further enhancing its capability to manage complex projects efficiently across different geographical locations.

The Judge Group, a global services firm, has significantly increased its operational efficiency by adopting Box Extract and Box AI Studio, leading to an 80% boost in speed. This transformation has enabled the company to drastically reduce contract review times from six hours to one hour and streamline compensation document processing from up to 60 days to just 10 hours. By leveraging Box Extract to convert unstructured data into structured metadata, the Judge Group automates workflows, reduces manual data entry, and enhances document accessibility across its 26 offices. Custom AI agents perform multiple checks on contracts, contributing to faster and more accurate processing. This shift has not only saved hundreds of work hours but also empowered employees to focus on tasks that require human expertise. The firm's successful integration of AI and automation indicates a cultural shift where employees embrace technology as a tool to enhance, rather than replace, human work. Looking forward, the Judge Group plans to further integrate Box with other platforms like ServiceNow to enhance procurement workflows and utilize AI for proposal generation, demonstrating a forward-thinking approach to intelligent content management.

State and local government agencies are exploring the potential of AI to enhance operational efficiency and service delivery, yet face challenges in moving AI pilots from experimentation to production, as highlighted at the Box State and Local Government Summit. The summit underscored the importance of unifying content to leverage AI's potential, emphasizing that agencies often struggle with fragmented data systems that impede AI effectiveness. Participants advocated for starting AI projects with real operational problems rather than mere technology experiments, focusing on high-impact areas such as automating paper forms and enhancing document intake processes. Demonstrations at the summit showed AI's role in supporting, not replacing, human decision-making, particularly in claims management and eligibility determinations, where AI provides data-driven insights while preserving audit trails. The discussions also stressed embedding security and governance into AI initiatives to protect sensitive data, with case studies like the Georgia Department of Community Health showcasing successful integration of AI with rigorous security protocols. Finally, the City and County of Denver's approach to Intelligent Content Management demonstrated the benefits of standardized, scalable implementations that meet diverse agency needs while fostering innovation and improving public service outcomes.

BTS, a global professional services firm, has observed a common challenge in AI adoption where access to AI tools does not necessarily lead to true operational change. They emphasize the importance of integrating AI directly into existing workflows rather than adding new tools, as demonstrated by their custom AI-driven workflow that streamlines client contact monitoring and communication, enhancing efficiency and CRM data quality. BTS uses Box as a centralized content layer to support AI initiatives, ensuring data quality and governance while fostering innovation within secure boundaries. Their approach to AI measurement focuses on operational metrics to assess the real impact of AI on business processes, highlighting the importance of embedding intelligence directly into daily work to achieve scalable and valuable transformations.

Box, Inc., a leader in Intelligent Content Management, showcased the transformative use of its AI-powered platform by state and local governments at its annual State and Local Government Virtual Summit. The event highlighted the platform's role in enhancing public-sector operations through secure and compliant content management, featuring use cases from the City and County of Denver, the Texas Department of Motor Vehicles, and the Los Angeles City Employees’ Retirement System (LACERS), among others. These organizations leverage Box's technology for various applications, such as automating workflows, ensuring compliance, and improving service delivery, all while addressing the challenges of budget constraints and legacy systems. Box's platform integrates with numerous business applications, providing scalability, security, and compliance with standards like CJIS, HIPAA, and FedRAMP. The summit emphasized Box's commitment to delivering AI-driven insights and efficiencies, facilitating better mission delivery, and reducing IT expenses for public sector agencies.

AI is revolutionizing cybersecurity by enabling continuous adversarial validation, a departure from traditional, episodic penetration testing, which no longer suffices in a rapidly evolving threat landscape. As AI accelerates the ability of attackers to find and exploit vulnerabilities, the need for a dynamic approach to security becomes imperative. Continuous adversarial validation leverages AI to simulate attacker behavior, continuously probing and mapping the entire attack surface, including applications, APIs, identity systems, and infrastructure. This model allows for real-time evaluation of exploitability as systems change, ensuring that security efforts focus on actual risks rather than theoretical vulnerabilities. By combining AI-driven testing with human expertise, organizations can maintain an up-to-date understanding of their security posture, allowing human penetration testers to concentrate on complex and novel attack paths that require deep adversarial thinking. The shift from calendar-driven to signal-driven security validation ensures that defenses are tested in alignment with the current system state, enhancing the ability to proactively manage security in the face of increasingly sophisticated AI-powered adversaries.

Georgia's Department of Community Health (DCH) is transforming its public health infrastructure by adopting Box as its Intelligent Content Management platform to enhance security and efficiency in managing Medicaid workflows for over 10 million residents. This initiative is part of a broader effort to strengthen rural healthcare access and ensure sensitive medical data remains secure and accessible across the state. By integrating Box AI and Box Shield Pro, DCH is automating data classification and protection, replacing outdated manual processes, and enabling more efficient handling of Protected Health Information (PHI). The partnership with Box, which complies with FedRAMP and HIPAA standards, allows the agency to effectively manage cybersecurity risks and streamline operations, empowering staff to focus on their primary mission of serving the community. This modernization effort, backed by substantial federal funding and involving collaboration with multiple partners and vendors, aims to set a precedent for other government agencies seeking to enhance their operational resilience and data security through advanced technology solutions.

AI coding agents have become proficient at writing code and handling complex tasks, especially when equipped with domain-specific knowledge through tools like Codex skills. A Codex skill is a structured knowledge package that enhances an AI agent's capability to work with specific platforms by providing essential reference materials and workflows, thereby mimicking the expertise of a seasoned developer. This post outlines the function and benefits of Codex skills, using the Box Content API as a case study, showcasing how a skill can streamline processes such as file organization and content classification through efficient use of platform-specific tools like Box AI. The skill employs a routing table to ensure relevant documentation is accessed as needed, maintaining focus and accuracy while reducing unnecessary API usage. Additionally, the importance of precise guardrails in guiding AI behavior is emphasized, demonstrating how iterative testing and refinement can optimize agent performance in executing tasks like document classification and bulk operations.

As organizations rush to implement AI, they face a significant challenge: the quality of their AI systems heavily depends on the integrity of their data, which is often compromised by redundant, obsolete, and trivial (ROT) data. Mike Carlino from Deloitte highlights that around 40% to 50% of enterprise information is ROT, resulting from poor information-sharing practices and inadequate data governance. This data quality issue, known as "Garbage In, Garbage Out" (GIGO), can hinder AI initiatives, turning promising AI pilots into failures when scaled to production. To overcome this, enterprises must establish robust data governance frameworks, centralize content management, and ensure clear data categorization and integration. Industries like financial services, which deal with structured data, have an advantage in applying AI effectively, whereas others must work harder to clean their data and implement governance to achieve reliable AI outcomes. The key to success lies in starting with controlled, predictable data flows and committing to ongoing data management and governance.

Traditional security architecture and design reviews are becoming obsolete due to the increasing complexity of modern systems and the accelerating pace of product development, which outpaces the capacity of small security architecture teams. The emergence of agentic development, where AI agents handle coding, review, and testing tasks, necessitates a shift from human-centric security processes to machine-consumable security intelligence. This transformation involves integrating security guidance directly into the development lifecycle, ensuring that security considerations are embedded at the design stage rather than as an afterthought. By automating tasks like STRIDE threat modeling and enabling agents to consume structured security artifacts, the process becomes scalable and efficient, reducing the cognitive load on human security architects and allowing them to focus on complex, creative security challenges. Additionally, this approach creates a dynamic feedback loop where security intelligence continuously adapts to changes in design, leading to more robust and responsive security measures that align with the rapid pace of modern software development.

The new version of the Box CLI has been launched to facilitate secure and programmable interactions with Box content, catering to both developers and AI agents. Originally designed to automate tasks in the terminal, the updated CLI now supports advanced AI tools like Claude Code, enabling users to perform complex operations such as file retrieval, metadata application, and document transformation with ease. To get started, users can install the CLI, authenticate via OAuth or a Platform App, and utilize the command line for a variety of tasks. This includes automating workflows like uploading files, extracting metadata, and managing documents, all of which are made simpler for AI agents to execute. The new CLI also supports document signing, watermarking, and other functions, enhancing the automation capabilities available to developers and agents.

In November 2026, AI agents are autonomously negotiating transactions without human intervention, exemplified by the viral platform Moltbook, which showcased both the potential and risks of such technology. While Moltbook demonstrated that AI agents can autonomously coordinate and adapt, it also highlighted significant security flaws, as evidenced by data breaches and prompt injection attacks. The incident underscored the importance of robust security architecture, verified identity, comprehensive auditability, and clear liability frameworks for enterprise AI systems. As the industry moves towards agent interoperability, standardized protocols and secure governance frameworks become essential to balance innovation with risk management. Box has proactively addressed these concerns by implementing the Agent2Agent (A2A) protocol and the Box MCP Server, ensuring authenticated, logged, and scoped agent interactions, thereby reducing the reliance on manual processes and enhancing efficiency in enterprise operations.

Autonomous agents are emerging as powerful tools for enterprises, capable of executing complex tasks independently, yet presenting challenges in terms of security and governance. To address these challenges, Box is collaborating with NVIDIA to create an infrastructure that allows AI agents to access and interact with unstructured data securely, maintaining essential compliance and privacy controls. This partnership involves the development of the NVIDIA Agent Toolkit and the open-source stack, NVIDIA NemoClaw, which facilitate the safe operation of AI agents within the Box file system through the OpenShell runtime. These agents are able to perform tasks such as document generation and data extraction while adhering to strict security policies integrated into the Box environment, ensuring that all actions are logged and compliant with enterprise standards. As organizations increasingly deploy autonomous agents, the need for a robust, secure content management system becomes crucial, and Box positions itself as the foundational layer that harmonizes AI agility with enterprise governance, enabling the seamless and safe deployment of agents at scale.

Modern vulnerability management faces significant challenges due to increasing CVE volumes, complex software supply chains, and limited resources, with the urgency heightened by AI reducing the time from vulnerability disclosure to exploitation. Despite the vast number of vulnerabilities, only a small percentage are actually weaponized, prompting a shift from traditional volume-based models to risk-based approaches that prioritize the vulnerabilities posing real threats. At Box, AI is leveraged to transform vulnerability management from a reporting system into a risk-reduction engine, focusing on context-aware prioritization and intelligent remediation. This model emphasizes reducing exposure conditions, implementing AI-powered automation for proactive detection, and providing developers with real-time, context-aware solutions to prevent vulnerabilities from reaching production. The strategy aims to shrink the attack surface by addressing vulnerabilities earlier and reducing recurring exposure through disciplined engineering practices. However, challenges remain in data quality and ensuring transparency in AI-driven risk scoring, necessitating continuous model refinement to maintain trust and operational stability.

AI agents can efficiently handle tasks involving documents stored in cloud storage by utilizing a combination of tools and structured processes. The described system involves a team of Claude agents using the Box CLI to interact with files in Box, enabling them to search for documents, extract data, and compile a summary spreadsheet to share with a tax preparer. Starting with a folder containing tax documents, the process involves several agents, each with a specific role: the Explore Agent classifies documents, the Extract Agent extracts data, and the Excel Agent compiles and shares the spreadsheet. The Plan Agent orchestrates these tasks, breaking down a high-level goal into manageable tasks and coordinating the execution. Each agent operates independently, equipped with a system prompt and a set of tools, and the user retains control over task execution through a real-time UI. The system effectively integrates the Claude Agent SDK and Box CLI commands to automate and streamline document management and data processing tasks.

Owens Corning, a global building materials manufacturer, has transformed its operations by integrating AI into its traditional industry practices, fostering a culture of innovation and experimentation among employees. Under the leadership of Senior Vice President and CIO Annie Baymiller, the company shifted from top-down mandates to a more grassroots approach, encouraging engineers, safety managers, and scientists to explore AI applications that improve efficiency and decision-making. This transformation led to the digitization of 85 years of R&D data, creating a searchable AI-powered knowledge base that accelerates innovation by preventing redundant research. In terms of safety, Owens Corning uses AI for predictive risk management, analyzing historical safety data to foresee and mitigate potential hazards. The company also established a balance between innovation and governance through an AI Community of Practice and a Responsible AI Group, ensuring safe and ethical AI experimentation. Owens Corning’s approach, characterized by a "breadcrumb strategy" that celebrates small victories and makes AI accessible, serves as a model for traditional industries looking to adopt AI effectively, emphasizing the importance of data foundation, clear boundaries, and storytelling to drive transformation.

Trócaire, a prominent international development charity, has significantly improved its operational efficiency and compliance by transitioning from a fragmented data management system to a unified cloud-based platform using Box. This transformation involved consolidating scattered local servers and multiple data storage methods into a single secure system, enhancing the organization's ability to manage 1.5 TB of donor-related content with reliable compliance traceability through a Salesforce integration. The shift was not merely technological but also cultural, requiring a change in mindset across the organization to adapt to new ways of working and improve collaboration in challenging environments such as conflict zones and disaster-hit areas. The mobility and accessibility provided by Box have transformed travel time into productive time for Trócaire’s mobile workforce, while the integration of Box AI and Box Shield is laying the groundwork for future innovations in data management and security. This strategic shift has not only streamlined operations but also positioned Trócaire to leverage emerging technologies, ensuring the organization can continue to deliver aid effectively and securely, regardless of location or circumstances.

Box is revolutionizing its cybersecurity approach by developing an AI-driven detection ecosystem to address the inefficiencies and delays inherent in traditional detection engineering. The new system aims to enable near real-time detection by leveraging AI to automate the ingestion, understanding, and translation of threat intelligence into actionable detection logic, significantly reducing the time from intelligence alert to detection deployment. This approach shifts the focus from manual and reactive processes to proactive and continuous detection coverage, improving the quality and relevance of security insights while freeing analysts to concentrate on higher-value tasks such as adversary chain analysis and detection stress testing. Despite challenges like translating narrative threat reports into precise detection logic and ensuring rigorous validation, Box anticipates that this integrated system will enhance detection engineering throughput and signal quality without compromising rigor. The initiative underscores the necessity for security operations centers to adapt to the fast-paced evolution of cyber threats and remains committed to overcoming integration complexities and ensuring robust feedback loops for continuous learning and improvement.

OpenAI has released GPT-5.4, which shows notable improvements over its predecessor, GPT-5.2, by enhancing overall extraction accuracy from 72% to 78%, particularly benefiting enterprises that depend on AI for data extraction, analysis, and workflow automation. This new model excels in tasks requiring multi-step calculations, deep reasoning, and inference from complex documents, with improved performance across various industries and document types, such as clinical data, legal agreements, and regulatory filings. GPT-5.4 also demonstrates significant gains in core extraction categories, such as handling fields with heavy calculations, executing multi-step reasoning tasks, and extracting numerous fields from extensive documents. The model is particularly adept at reducing errors and omissions, providing more reliable data extraction in specialized content areas like healthcare, legal, and energy, and is now available in Box AI Studio for building custom AI workflows.

AI coding agents have evolved beyond mere code suggestions to perform complex actions such as running commands, calling APIs, and making decisions across workflows, with the Model Context Protocol (MCP) allowing them to interact with external platforms like Box. This capability raises concerns about data security and governance, which the Box Plugin for Cursor aims to address by using a lightweight package with three components: a connection protocol (mcp.json), skill instructions (SKILL.md), and mandatory safety rules (box.mdc). The plugin ensures AI agents have access to Box tools, know how to use them responsibly, and adhere to safety guardrails, thereby allowing developers to automate tasks like document retrieval and summarization without manual intervention. The plugin's design, which uses natural language to encode instructions and constraints, represents a shift in how developer tools are built, emphasizing safe and competent interaction with enterprise content.

In a discussion between Box CTO Ben Kus and Braintrust CEO Ankur Goyal, it is proposed that AI excels more in validating responses than in generating content, a concept dubbed the "grading paradox." This insight suggests that the real advancement in AI comes from enhancing its ability to recognize good answers, similar to how humans find essay grading easier than writing. This shift from deterministic to non-deterministic AI requires a fundamental change in software development, focusing on building evaluation frameworks that leverage AI's grading capabilities. By prioritizing evaluation sets over models, enterprises can better manage AI's inherent unpredictability, transitioning from impressive demonstrations to reliable production systems. Emphasizing measurement over model selection enables teams to systematically capture and quantify what constitutes a "good" response, thereby making AI applications more effective and reliable.

Vornado Realty Trust, a leading Manhattan commercial real estate company, significantly improved tenant services and operational efficiency by utilizing Intelligent Content Management from Box, incorporating AI capabilities. By deploying Box AI and Box Hubs, Vornado reduced tenant inquiry response times by 30%, streamlined lease and project management, and enhanced security and collaboration across its properties. The integration of Box AI allows for automated responses to common tenant inquiries and assists in project management by automatically extracting action items from meeting transcripts. This strategic use of technology not only optimized internal workflows but also maintained strict access controls, ensuring secure document sharing with tenants and third parties. Looking forward, Vornado plans to deepen the integration of Box’s AI-driven workflow tools to automate document-heavy processes like invoicing and lease administration, setting a new standard for efficiency in the competitive NYC commercial real estate market.

Justice and public safety organizations are increasingly dealing with vast amounts of digital evidence and sensitive data, necessitating advanced management solutions. Box, a leading Intelligent Content Management platform, is addressing these challenges by providing a unified, secure platform that integrates diverse data sources and alleviates the issues of vendor lock-in and fragmented workflows. With Sabih Khan leading the strategy, Box is focusing on delivering industry-specific solutions and AI-driven workflows tailored to justice and public safety (JPS) agencies, automating critical tasks such as transcription and discovery preparation. This approach aims to enhance productivity, streamline collaboration, and enable agencies to meet compliance demands while improving citizen services. Box's commitment to dismantling technical silos and introducing standardized digital evidence management solutions positions it as a critical partner for agencies seeking to modernize their operations and better serve their communities.

Box has been awarded the G2 Award for Best Content Management Software, a distinction based on over 6,800 customer reviews and a 4.3-star rating, underscoring its reputation as a leader in the content management field. This recognition is notable as it is derived entirely from user feedback, highlighting Box's effectiveness in managing, securing, and facilitating collaboration on critical content. Customers particularly appreciate Box's AI-powered workflows, which streamline processes and provide insights, as well as its robust integrations with tools like Microsoft 365, Slack, and Salesforce. Users also praise the platform's ease of use, secure collaboration capabilities, and comprehensive file control settings. This accolade emphasizes the trust users have in Box and its commitment to delivering secure and intelligent content management solutions, especially as organizations increasingly prioritize AI-driven strategies.