It’s no mistake that architecture is a word used in the tech world to describe the design and structure of a computing system. Like when architecting a physical building, it’s critical to lay a strong foundation and then use the right raw materials and techniques as you build. In this new era of AI, a solid content foundation is particularly important, and so is the way you build with it.
Marx Okubo is a national architecture and engineering consulting firm known for delivering detailed building assessments and project plans. With eight offices across the US, the firm’s employees manage thousands of lengthy narrative reports created for clients looking to acquire or sell property. These reports can be anywhere from 20 to 200 pages long.
Before adopting Box, extracting actionable insights from deep within these massive documents was a “nonstarter,” according to Adam Wong-Brooks, part of the technical team at Marx Okubo. It was such an onerous task that Wongs-Brooks says, “If we were to try to accomplish it with employees in the company, it would be a prohibitively costly or time-consuming process.”
Now, with Box AI and especially Box AI Studio, he says, “What was previously a nonstarter is now possible.”
Key takeaways from Marx Okubo’s story for any enterprise:
- Unlock value from unstructured data by applying metadata extraction APIs integrated seamlessly into existing environments
- Empower users at all levels through easy-to-use interfaces like custom AI agents that reduce reliance on specialized tech skills
- Maintain rigorous security controls while enabling collaboration — even under demanding regulatory requirements — with flexible platform features like geofencing and watermarking
Valuable information buried within ~10K of unstructured narrative reports
Unlike structured data neatly formatted in rows and columns, the over 10,000 detailed reports Marx Okubo contends with are in an unstructured narrative format. Manually extracting information from these documents would have been "prohibitively cost or time consuming,” according to Wong-Brooks. The challenge involved accessing valuable information buried within thousands of narrative reports scattered across a sprawling folder environment.
By implementing automatic metadata extraction via the Box API, his team developed a sophisticated workflow that could programmatically query relevant files, extract structured data from unstructured content, and apply those metadata values back to the reports. This transformation created a structured framework for interacting with previously inaccessible information.
The process allowed them to surface insights that would have been buried in narrative reports, enabling both internal teams and clients to make better-informed decisions. As Wong-Brooks noted, “I would say it's an infinity difference, because prior to Box AI, we weren't able to do it at all.”
I would say it's an infinity difference, because prior to Box AI, we weren't able to do it at all.
Infinity difference when put to the test
By leveraging Box AI Studio to create a custom Materials Agent that reviews monthly structural observation reports (often 90+ pages each), Marx Okubo slashed review times dramatically. The agent automatically identifies materials mentioned in reports, summarizes conditions clearly, and flags issues — all within minutes (instead of hours spent reading manually).
This automation frees technical staff to focus on higher-value analysis rather than repetitive document parsing, while ensuring no critical detail slips through unnoticed. When one financial institution client needed data on electrical capacity across industrial buildings nationwide, Marx Okubo could efficiently extract specific information from relevant reports and export it to CSV format for analysis.
This automated extraction process transformed what was previously impossible into a streamlined workflow. Wong-Brooks emphasized the dramatic impact of this capability, describing it as "an infinity difference" compared to their previous limitations, enabling them to unlock insights that were previously impossible to access at scale.
The extra complexity of working with sensitive data
Another layer of complexity Marx Okubo contends with is the challenge of working with sensitive data, particularly because the company collaborates with a government agency that imposes unique and strict data requirements. “This was a big deal,” says Matt Hoey, AVP of Technology, “but we were able to stay within our familiar instance of Box and build out a whole new folder structure so we could take advantage of things we didn’t need before.”
Box provided a comprehensive solution that allowed Marx Okubo to maintain its familiar environment while implementing advanced security measures tailored to stringent requirements. Rather than investing in additional providers or consulting services, the team leveraged Box's robust security features — including geofencing to restrict access by location, watermarking to protect document integrity, and download prohibitions to prevent unauthorized distribution.
If we had gone with another cloud provider not innovating like Box, I don’t know if we could have built these solutions ourselves.
This security-focused approach gave Marx Okubo the confidence to handle highly regulated information while maintaining compliance, all within their existing Box instance, and without disrupting established workflows or requiring additional technology investments. Wongs-Brooks confirms, “If we had gone with another cloud provider not innovating like Box, I don’t know if we could have built these solutions ourselves.”
Beyond content access and security
Today, Marx Okubo takes advantage of Box for multiple Intelligent Content Management solutions.
- Box Hubs: Centralized portals organize project-specific content, accessible internally and externally
- Box AI Studio: Custom-built AI agents automating tedious tasks such as summarizing long materials-testing reports
- Box AI Extract Agents and Extract Agent APIs: Programmatic extraction turning unstructured narratives into searchable structured data fields
In terms of metadata extraction, Wong-Brooks explains, “We adapted scripts from the Box developer library so that we could extract metadata not just against one folder but across our entire file environment. It became a matter of testing accuracy after that.”
Internally, they also use Box Notes (combined with querying functionality) so any employee can easily search transcripts or documents without needing database expertise — a true democratization of knowledge access across departments.
The AI-powered future for Marx Okubo
Marx Okubo’s journey shows how combining trusted cloud storage with powerful AI-driven automation transforms traditional workflows plagued by complexity and scale challenges.
