For financial services companies, the competitive advantage has always been hiding in plain sight: critical data locked inside documents no one has time to read. AI-powered data extraction is changing that.
Underwriting teams at real estate lender Barnett Capital used to routinely spend an hour reading each appraisal to find critical information like dates, terms, and figures. Now, they get answers to 35-40 standard questions in about a minute, regardless of how the appraisal is formatted.
At Valmark Financial Group, analysts used to pull just 4,000 data points annually from insurance policy illustrations — dense tables of data that required tedious manual review. Today, they easily extract over 240,000.
At USAA, a member-owned financial services company serving 13 million military families, IT leaders have been able to mine long IT contracts for answers to questions in seconds, giving them back time for more valuable strategic planning.
And at Mercer Advisors, an independent fiduciary financial advisory firm, files scattered across emails and folders created a lot of busywork for advisors whose time was better spent building client relationships. Within client onboarding alone, AI-enabled data extraction meant radically faster onboarding.
Four financial organizations. Four different challenges. One common thread: AI-powered data extraction with Box Extract made their invisible document data visible.
Automated data extraction layered on secure Intelligent Content Management
Locked inside financial services content is a goldmine of important data, it’s not always easy to find when you need it. Storage isn’t the crux of the challenge. Most companies have already digitized their documents and moved them to the cloud. The real challenge is finding, capturing, and understanding what’s buried within content — turning unstructured content into actionable information without manually reading every page.
In an industry where speed often determines who wins the deal, manual data extraction is becoming a serious competitive disadvantage. A 2025 BCG report stated that only a quarter of banks were using AI competitively last year, and that “winning in the coming era will require far more than isolated pilots or cautious upgrades.” BCG asserts that “Making AI work at scale requires rethinking the architecture itself,” and that’s what our four hero companies have done.
Barnett Capital, a Chicago-based family office lender, underwrites residential fix-and-flip loans and commercial triple-net lease financing — deals that require quick turnaround. Getting quick insight from large-volume unstructured files like appraisals and leases is the key to fast lending decisions.
To support its underwriting, Barnett centralized all its files on Box and integrated that content with Salesforce. The team was then able to apply Box Extract to automatically, and securely, extract the relevant metadata from files during the underwriting process. “By applying metadata extraction,” says Barnett Capital Director Dan Nikolich, “a task that used to take an hour now takes about a minute.”
Faster knowledge management as financial firms grow
Valmark Financial Group serves roughly 400 financial advisors nationwide, acting as their back office for complex financial needs. Insurance illustrations (the dense documents that project policy performance over decades) are "a table with lots and lots of rows of data," explains Geoff Moore, Valmark's CIO.
Before AI extraction, he says, "It was too time-consuming to load all that data, so we would just load a single row, and the next year, come back and lay another row over it." Decades of valuable information sat untouched because no one had time to manually extract it.
Valmark solved the problem by building on Box as a secure foundation they could layer additional IT capabilities on top of. With a secure, compliant file system for its documents, Valmark could use Box Extract to automate metadata extraction from complex insurance illustrations and save analysts days of manual work.
A shift in applied human capital
By automatically pulling structured data from unstructured documents — things like contract terms, expiration dates, signed status, and key figures — AI extraction can drastically change how financial services companies interact with their content. Even if these pieces of information are inconsistent and formatted in wildly different ways, AI can parse the correct information and turn it into structured metadata. That metadata is then searchable, reportable, and actionable. What once required a human to read and interpret can now happen automatically, at scale.
At USAA, automated data extraction has helped IT teams manage vendor relationships with speed and precision. "I use Box AI to quickly retrieve SLOs, expiration dates, and renewal terms without having to read the entire document," says Otis Avworo, Director of IT for Collaboration Technologies.
For an organization with over 70,000 employees and a century of institutional knowledge, the ability to surface specific information on demand has been a major time-saver. Avworo says, “I now have more time to dedicate to strategic planning, ensuring system uptime, and providing our workforce with the technology they need to effectively serve our members.”
When organizations like USAA and Valmark are able to eliminate the hours spent finding information, their people can focus on using it. Underwriters can analyze deals instead of hunting for data points. IT leaders can focus on strategic planning instead of contract review. Paralegals can support higher-value work instead of searching through email archives.
Human analysis aided by automated data extraction
Onboarding new clients is critical to business growth — and can be one of the most tedious administrative activities for advisors, who weed through a lot of documentation before accepting and onboarding a client. At Mercer Advisors, that activity includes reviewing complex tax returns, which traditionally took an advisor two weeks. By embedding intelligent automation, Mercer was able to speed that process up. Today, new clients upload their documents to a portal, which shuffles the files directly into the appropriate Box folders.
Mercer President Daniel Gorvitch says, “We went from two weeks down to five minutes getting quotes based on extracted tax return info — that fundamentally changes our sales process.”
Quantifiable productivity gains aside, the real value of automated data extraction is freeing up Mercer’s advisors to eliminate bottlenecks and focus on the analytical and relationship-building aspects of their roles. As Gorvitch puts it:
It’s not just about having cool tools; it’s about transforming how you deliver value.
Creating a secure, governed content layer
For financial services companies, any AI solution must meet stringent security and compliance requirements — FINRA and SEC 17a-4 record-keeping mandates, Dodd-Frank and CFPB obligations, data privacy frameworks like GLBA, CCPA, and GDPR. Documents in this industry inherently contain sensitive personal information, proprietary deal terms, and regulated data that cannot be exposed to risk.
For data extraction, platform matters as much as capability. When extraction runs inside a secure content management system, with permissions, audit trails, and compliance controls already in place, organizations can adopt AI without creating new security vulnerabilities. The intelligence layer inherits the governance of the underlying platform.
When every piece of sensitive content — from loan packages and KYC files to contracts and client communications — is governed in one place, you stop worrying about what's falling through the cracks. Retention policies, legal holds, and full audit trails built directly into the platform mean your risk, compliance, and audit teams always know where content lives, who touched it, and why. Permissions-aware AI ensures agents and users can only access what they're authorized to see — no shadow AI, no data leakage. The result is a content foundation that keeps you audit-ready and in control, no matter how fast your AI initiatives move.
Compounded benefits of automated AI extraction
When document processing accelerates, the benefits compound across the organization.
Faster decisions. At Barnett Capital, the speed improvement has been dramatic. By centralizing deal files in Box, integrating with Salesforce, and applying AI extraction, Barnett has built a workflow where metadata automatically populates and triggers downstream processes. At Mercer Advisors, too, onboarding new clients is a much quicker process with information extracted from tax returns quickly.
Reduced headcount pressure. Now, organizations can scale operations without proportionally scaling teams. Growth doesn't automatically mean hiring more people to read more documents. "This puts time back in our day," says Nikolich, at Barnett. "Even with our business growing, we haven't needed to hire more people."
Better data hygiene. Extracted metadata can be used to answer immediate questions, but it can also populate searchable fields that improve discovery across the entire content lifecycle. One commissions team member at Valmark avoided two full days of monthly manual entry after switching vendor reports from PDF feeds into automated extraction with OCR capability.
Improved job satisfaction. When staff shift focus away from repetitive extraction tasks toward higher-value analysis and client service, engagement improves.
What comes next in financial services
Companies like Barnett, Valmark, USAA, and Mercer aren't stopping at basic extraction. They're building workflows that trigger automatically when metadata is applied. They're creating self-serve portals where employees can ask questions directly across thousands of documents. They're piloting AI agents that replicate expert knowledge so teams can get instant answers without bottlenecks.
USAA plans to migrate legacy document stores to Box, scale metadata extraction across the enterprise, and leverage compliance and retention features to govern the content lifecycle. The vision is clear: "A single, secure, AI-powered content platform that works as hard as the teams it supports,” says Avworo.
At Valmark, Moore says, “Box isn’t just our document storage provider anymore. They’re our long-term AI partner helping us reimagine what’s possible.” The company plans to leverage new Box AI features and encourage employees to experiment with extraction prompts to further unlock data.
AI automation, built on accessible data, is the future for financial services companies.
Learn more about agentic data extraction for smart process automation with Box Extract. Or visit Box for financial services to find out how your organization can power seamless client experiences with AI.
