Workday Vice President of Evisort Jerry Ting breaks down how AI-powered document intelligence, when built with real data and domain expertise, can transform enterprise workflows at scale.
Ghadeer Redler
Head of Multimedia, Thought Leadership & Customer Stories
Workday
Listen to this video in audio format on Spotify.
Every time a company hires someone, closes a deal, or pays a vendor, it creates a paper trail—contracts, offer letters, invoices, approvals.
Now, multiply that by thousands. Across departments. Across systems.
Most of it is still handled manually. People spend hours hunting for the right version, copying clauses into spreadsheets, checking compliance line by line. The work is slow, error-prone, and draining.
That’s the problem document intelligence is built to solve.
In this AI Masterclass, Vice President of Evisort at Workday, Jerry Ting, shares what it really takes to move from scattered documents and reactive workflows to AI-powered document intelligence that drives real, repeated outcomes. In his deep dive of document intelligence, you’ll also learn why most AI fails in the wild, how to train models on messy real-world data, and what it looks like when enterprise organizations build together.
Report
Document intelligence is the use of AI to extract, analyze, and act on the data hidden inside business documents. These aren’t always clean rows in a spreadsheet. They’re often unstructured, nuanced files like contracts, offer letters, vendor agreements, policies, and invoices.
Most enterprise operations, especially in legal, HR, and finance, depend on these documents to function. Every employee onboarding, vendor payment, or deal closing creates a paper trail that someone has to manage, interpret, and track.
Traditionally, this is done manually. A team member downloads a PDF, opens it, reads through line by line, copies what matters into a system, and flags anything unusual. It’s time-consuming, error-prone, and disconnected from the rest of the business.
Document intelligence changes that. With the right AI, these files become searchable, sortable, and actionable. Data gets extracted automatically. Risky language can be flagged. Terms can be compared. Statuses can be tracked.
This shift reduces operational drag and unlocks faster, more accurate decision-making. Document intelligence turns documents from blockers into drivers of efficiency and insight.
Document intelligence turns documents from blockers into drivers of efficiency and insight.
Despite widespread digital transformation, most document-heavy workflows still rely on manual labor. An HR rep might spend hours scanning onboarding documents. A paralegal might manually track renewal dates in a spreadsheet. A finance analyst might review invoice terms by hand to ensure compliance.
This approach isn’t just inefficient. It creates risk. Manual reviews miss things. Processes stall when one person is out. And high-value employees are stuck doing low-value work. That friction builds up across departments. It slows hiring, delays payments, and introduces errors. Over time, it adds cost, hurts morale, and limits scale. AI offers a way out, especially when it’s grounded in reality.
AI tends to look amazing in a demo. The model extracts data instantly, the results are perfect. But that’s because the data is clean, curated, and ideal. The real world isn’t always like that.
Many documents are scanned sideways, and contain handwritten annotations, inconsistent formatting, embedded tables, legal jargon, and visual noise. They don’t follow a script. And they certainly don’t care if your model’s confidence score is 94%.
That’s why AI built on synthetic or test data can fall apart when deployed, Ting explains. It wasn’t trained on the kind of messy, unpredictable inputs it will actually encounter.
To succeed, AI must be exposed to the same variability it will face in production. That means using real customer documents, with all the noise, inconsistency, and edge cases that come with them, to train and test models. When AI succeeds under those conditions, it earns trust, which leads to sustained adoption.
AI can’t work in a vacuum. Understanding the technical side is only half the job. You also need to understand the work that the AI is meant to support.
Legal teams aren’t just scanning documents, they’re assessing risk. HR professionals aren’t just processing forms, they’re ensuring compliance and protecting privacy. Finance teams aren’t just tracking invoices, they’re managing cash flow and audit trails.
That’s why these domain experts like lawyers, HR leaders, accountants, and compliance officers need to be embedded in the AI development process. They help define what matters, what’s risky, and what a helpful output looks like in the real world.
Ting, who is the founder of Evisort, built the company around this very principle. Many of Evisort’s product leaders are former practitioners who work side by side with technologists to co-design solutions that reflect the real stakes of enterprise work. This collaboration leads to AI that not only functions technically, but delivers value in context.
AI can’t work in a vacuum. Understanding the technical side is only half the job. You also need to understand the work that the AI is meant to support.
Documents rarely stay in one department. A single contract might originate in sales, be negotiated by legal, approved by finance, and archived by procurement. Applying AI to any one stage of that lifecycle without coordinating the others is a short-term fix that creates long-term fragmentation.
That’s why document intelligence must be architected with interoperability and alignment in mind. It should facilitate shared understanding across roles, not reinforce silos.
Operational alignment also depends on governance: who owns the AI system, how updates are managed, and how users are supported. Without clear ownership and collaboration across teams, even the best AI model will underdeliver.
Enterprise-ready document intelligence is not a point solution. It’s a shared capability layer that integrates across systems, stakeholders, and decisions.
Like any new technical implementation, scaling document intelligence isn’t just about technical readiness. It’s about strategic integration. AI needs to be embedded into workflows, reinforced through change management, and monitored for performance over time.
This means asking clear questions: Where will this AI system add the most value? What data will it rely on? Who is responsible for tuning it as business needs evolve? What happens when it gets something wrong? The goal can’t just be rapid adoption. It should be sustained, trusted use. That requires planning beyond rollout. It requires systems that are resilient to change, and vendors who are true partners in evolution.
For leaders thinking beyond experimentation, document intelligence offers a powerful way to improve speed, reduce risk, and scale insight across the business. Here’s what to remember:
Real-world documents are messy. Your AI needs to be ready for that.
Training on test data won’t always cut it. Use actual documents from real workflows.
Practitioners should shape the product, not just use it after the fact.
AI success depends on trust, not just performance, and trust is earned by delivering consistent, explainable results.
Buy-in across departments is critical. Document intelligence is most valuable when it spans the enterprise.
What Ting outlines for document intelligence applies across the AI landscape: success doesn’t come from flashy features, but from systems that reflect how people actually work. Whether you're automating documents or deploying AI elsewhere in the enterprise, the path to value is the same: reliable data, domain depth, and cross-functional alignment.
Report
Pressure for finance leaders to invest in AI is at an all-time high, but where are CFOs spending their technology budget? Read on to learn more about the potential impact of widespread AI implementation from 2,355 senior business executives.