AI Powered Document Extraction

What if the information you need most is buried in handwritten notes, scanned PDFs, or outdated systems—but you still need it structured, accurate, and scalable across tens of thousands of records? Whether you're analyzing court waivers, social services forms, licensing records, or housing applications, the raw materials are often the same: fragmented documents in inconsistent formats. Solving that bottleneck doesn’t just speed up workflow—it unlocks entirely new ways of understanding the data you already have.

Example Problem

Oklahoma’s HB 2259 was created to reduce court costs for indigent Oklahomans by assessing their ability to pay. But key questions remain:

• Do these waivers result in meaningful financial relief?
• What is the total amount of court debt avoided?
• Does reduced debt lead to fewer failure-to-pay warrants or future charges?

Answering these questions requires structured, case-level data—yet the relevant details are buried in fragmented, scanned court documents with a mix of typed and handwritten text.

Our Solution

We’ve built a tightly controlled, AI-powered system that extracts and structures complex court data at scale.

• We retrieve relevant records directly from Oklahoma’s public court system using automated search tools.
• We use structured AI tools—guided by Pydantic schemas—to extract key information (balances, eligibility, court findings) from mixed-format documents.
• We store data in a graph database, making it easy to query like a spreadsheet while also mapping deeper relationships across cases, individuals, and outcomes.

Why This Matters

Beyond court records, this system offers a blueprint for structuring document-based data in any domain:

• Health: Extract data from handwritten intake assessments or care plans
• Education: Structure student-level records from scanned IEPs or counselor notes
• Workforce: Parse job training documents, evaluations, or subsidy applications

If your data lives in messy PDFs, DataWorks can turn it into something you can search, analyze, and use—at scale, across systems, and with validation built in.

Structured Event Tracking

What if you could automatically track key dates—hearings, renewals, deadlines—across dozens of systems and locations, without waiting for someone to manually check a website or pick up the phone? In every sector, missed dates create consequences: court warrants, lapsed benefits, missed interviews, expired licenses. The information is often public—but fragmented, inconsistently formatted, and hard to access at scale. We built a system that solves this for courts, but the same approach can power automated monitoring across public records—keeping individuals, teams, and systems on track.

Example Problem

Thousands of people miss court dates every year—not because they’re evading justice, but because they never receive clear or timely reminders. These missed appearances (failure to appear, FTA) often lead to bench warrants, arrests, and jail stays that could have been avoided.

The information needed to prevent this—next court dates, times, locations—is public, but spread across dozens of counties in inconsistent formats. For many individuals, family members, and case managers, keeping up with court dates is nearly impossible.

---

Our Solution

We’ve developed an automated system to extract, structure, and monitor upcoming court events across Oklahoma’s 77 counties.

• We pull next court events daily from public systems using custom-built tools that adapt to the format of each county’s data.
• Our AI structuring layer transforms raw, unstandardized listings into clean, queryable data—including defendant name, case number, hearing type, time, and location.
• All data is stored in a structured database, ready for integration into downstream tools such as text and email reminder systems.

Our system takes unstructured court listings across counties and clerks like this…

Monday, April 21, 2025 at 9:00 AM DISTRICT COURT ARRAIGNMENT DOE, JOHN CF Docket C

04-21-2025 CTPASS JUDGE Jerry Moses: Defendant present, not in custody and represented by Julie Johnson. State represented by John Johnson. District Court Arraignment passed to 4/21/25 at 9:00 AM in Room 406. Bond to remain; defendant recognized back.

…and transforms them into a structured record that can be passed to any downstream CRM or court reminder system.

				{
				  "Defendant": "John Doe",
				  "Case_Number": "CF-2025-XXXX",
				  "Next_Court_Event_Date": "2025-04-21",
				  "Next_Court_Event_Time": "09:00 AM",
				  "Next_Court_Event_Reason": "District Court Arraignment",
				  "Next_Court_Event_Docket": "Room 406",
				  "Court_Location": "Tulsa County Courthouse",
				  "Case_Status": "Pending" 
				}
			

---

Why This Matters

Court reminders are just the start. This type of system can apply anywhere structured scheduling matters—but the source data is a mess.

• Justice: Reduce failure-to-appear rates, automate diversion program tracking, or monitor warrants across jurisdictions.
• Workforce & Licensing: Track certification renewals, application deadlines, and compliance check-ins.
• Health & Benefits: Monitor Medicaid recertifications, WIC appointments, or eligibility hearings.
• Housing & Services: Keep clients and caseworkers informed about subsidy renewals, landlord hearings, or application windows.

Any organization that serves people navigating time-sensitive public systems can benefit from structured, real-time date tracking.

Comprehensive Case Modeling

Criminal court records hold the essential details that shape both individual outcomes and system-level insights—charges, sentencing, bond types, warrants, diversion participation, and more. But those elements live in many formats, scattered across scanned documents, docket minutes, and outdated court systems. There is no shared structure, no consistent language, and no reliable way to move this data across platforms or track it at scale—until now.

Support programs, researchers, and organizations investing in change can now access a stable foundation for service coordination, research evaluation, and insight analysis.

---

Our Solution

We’ve designed a legal data model that captures the full architecture of a criminal case—and built an AI-powered system to populate it automatically.

• We extract and structure dozens of case elements: disposition, sentencing type, custody, bond type, fines & fees, issued warrants, diversion participation, and court events.
• Every output is validated against a strict schema to ensure consistency and integrity—enabling integration with dashboards, case management systems, and research pipelines.
• Stored in a graph database, this model supports both individual case tracking and cross-cutting queries about sentencing patterns, bond trends, or program outcomes.

---

Structured Criminal Case Index

Example

---

Why This Matters

This isn’t just about structuring data—it’s about building a durable layer of infrastructure for the justice system.

• Programs can automate client case details, updates, and monitor legal events without manual follow-up.
• Funders and system leaders gain a unified view of how policies and interventions play out across populations and case types.
• Researchers get access to complete, clean data that enables evaluation far beyond what's possible with traditional court records.

Civil Case Intelligence

In civil litigation, the advantage often lies in what you can find—quickly. But the data needed to build timelines, spot expert witness patterns, or anticipate judicial rulings is scattered across filings, motions, scanned documents, and fragmented court systems.

Legal teams spend hours reviewing records just to answer basic strategic questions—and often outsource the work at high cost.

---

Our Solution

We’ve developed a system that extracts, structures, and organizes civil case data at scale—providing legal teams with the tools they need to work smarter, faster, and more strategically.

• Our AI models parse filings, dockets, motions, orders, and trial verdicts into structured case records — capturing outcomes, rulings, expert witnesses, timelines, and more.
• Using structured prompts and schema validation, we generate reliable case summaries and event timelines directly from raw documents.
• All data is queryable by judge, outcome, jurisdiction, and issue—revealing patterns in rulings, motions, and judicial tendencies over time.

Example Case Summary

On March 14, 2018, the plaintiff took their vehicle to the defendant for scheduled maintenance. The services performed included an oil and filter change, fluid level check, tire pressure adjustments, and a multi-point inspection. The defendant’s technician found that the tread on the plaintiff’s right rear tire was low. However, the defendant did not inform the plaintiff that the tire was unsafe. The plaintiff's car was returned to them on March 15, 2018. On March 25, 2018, the plaintiff was involved in a single-vehicle accident when the car hydroplaned during a heavy rainstorm.

The plaintiff contended that the accident occurred because the defendant failed to properly service the vehicle ten days earlier. Specifically, the plaintiff claimed the defendant was negligent in failing to inspect the tires and warn about the low tread. The plaintiff further alleged negligence in assigning the inspection to an employee who failed to properly notify them of the condition, and in not providing a written copy of the multi-point inspection results. The plaintiff also claimed breach of contract and breach of express and implied warranties, arguing that the maintenance and inspection services were not performed professionally. The plaintiff requested damages in excess of $75,000 for injuries, medical expenses, lost wages, and pain and suffering.

The defendant argued that the multi-point inspection was a courtesy and not a required service. They also claimed the plaintiff was distracted and speeding at the time of the accident. The defendant asserted there was no contractual obligation to provide inspection documentation and denied that the low tire tread caused the accident. They maintained they were not negligent and not responsible for the plaintiff’s injuries or damages.

Timeline of Events

• March 15, 2018: Plaintiff's vehicle undergoes maintenance at ABC Auto Services. Technician notes low tire tread.
• March 25, 2018: Plaintiff crashes due to hydroplaning. Plaintiff’s child was a passenger.
• 2018–2022: Legal preparation and evidence gathering.
• Feb 25, 2020: Plaintiff files lawsuit against ABC Auto Services and XYZ Auto Management Group.
• May 20, 2020: Defendants file responses, denying most allegations.
• Feb 22, 2021: Subpoenas issued for plaintiff’s employment and business records.
• Apr 28, 2021 – Aug 2022: Discovery requests, including attempts to schedule tire inspections.
• Aug 30, 2022: Plaintiff’s attorney requests VIN-based vehicle records.
• Dec 6, 2022: Motion to compel full discovery responses.
• May 8, 2023: Court orders defendant to comply with discovery.
• May–Sep 2023: Depositions of multiple service employees and experts.
• Oct 2023–Feb 2024: Ongoing discovery and correspondence.
• Dec 2023 – Feb 2024: Motions filed for summary judgment and discovery enforcement.
• Apr 12, 2024: Court grants defendant’s motion for partial summary judgment.
• May–Sep 2024: Trial preparation and objections exchanged.
• Oct 2024: Motions in Limine filed by both parties.
• Nov 4–7, 2024: Jury trial held.
• Nov 26, 2024: Jury finds ABC Auto Services 80% negligent; plaintiff 20% negligent.
• Dec 2024: Designation for appeal filed; defendants file for attorney fees.

Cast of Characters

• Alex Reynolds: Plaintiff, vehicle owner and driver at time of crash.
• Jamie Reynolds: Minor passenger in vehicle, child of plaintiff.
• ABC Auto Services: Main defendant, performed vehicle maintenance.
• XYZ Auto Management Group: Related business entity, co-defendant.
• Erin Matthews: Plaintiff’s lead attorney.
• Jordan Ellis: Co-counsel for plaintiff.
• Kai Walker: Defense attorney for ABC Auto Services.
• Taylor Brooks: Additional defense counsel.
• Casey Duran: Plaintiff’s engineering and tire expert.
• Robin Chen: Vehicle damage appraiser for plaintiff.
• Morgan Price: Technician who performed tire inspection.
• Drew Parker: Service department employee, trial witness.
• Jesse Lane: Auto technician, witness under subpoena.
• Sam Carter: Manager, witness under subpoena.
• Alexis Ford: Expert for plaintiff.
• Riley Kim: Additional service department employee and witness.
• Dana White: Former service advisor (no role in disputed appointment).
• Judge Morgan Bennett: Presiding judge over the case.
• Sydney Cross: Insurance adjuster for ABC Auto Services.
• RapidTech Solutions, Inc.: Third-party software vendor for repair tracking.

Why This Matters

• For Law Firms: Reduce time spent on manual case review, improve strategy, and gain a competitive edge with deeper visibility into courts, experts, and patterns.
• For Attorneys: Speed up research, improve accuracy, and make case information easier to access—freeing up more time for high-value, billable work.
• For Firm Leadership: Bring key research functions in-house, lower reliance on outsourced services, and invest in tools that compound in value over time.

About Us

At DataWorks, we believe that trustworthy, well-structured data is foundational to meaningful and lasting solutions. We design technology solutions specifically for complex data challenges—helping businesses, nonprofits, and public institutions alike turn messy, fragmented information into clear, actionable insight.

At the heart of our mission lies the DataWorks Commons—a growing initiative to make rigorously validated, high-impact datasets more accessible to the public. We envision the Commons as a shared resource for communities, researchers, advocates, and institutions working toward positive change.

Though still in development, the Commons reflects our deep commitment to accessible data and the public good—and we are actively seeking underwriters and partners who share our belief in its promise and potential.

Our work is grounded in values that drive both innovation and impact:

• Accessibility – We design tools and experiences that are intuitive, inclusive, and easy to navigate.
• Rigor – We validate every dataset to ensure accuracy, clarity, and real-world utility.
• Justice – We use data to illuminate disparities, support reform, and elevate community knowledge.
• Sustainability – We pair public benefit with durable infrastructure and responsible business models that support long-term stewardship.

At DataWorks, we’re building more than just tools—we’re building a trusted ecosystem where data works for everyone.

Brek Wilkins, Ph.D.

Brek has over 15 years of data analytics, technology development, operations, and leadership experience in higher education, the health tech industry, community non-profits, and grantmaking organizations. Brek obtained his PhD from OSU Center for Health Sciences and then served as Chief Technology Officer for a predictive analytics healthcare company, AmbreHealth Technologies. He supported operations, data analytics, and program evaluation projects at a Family & Children’s Services before serving as VP at Sway Medical, an innovative health technology company, where he built numerous consumer-facing data products. He most recently served as the Director of Analytics at ProsperOK where he provided data and technology support for partner investments and leadership at the George Kaiser Family Foundation.

Sam Johnson

Sam has over 30 years of experience designing large-scale technical architectures, building products, and managing teams. He graduated from Rice University with a Computer Science degree and worked with a systems integration consulting company for 5 years. He also was a VP at Goldman Sachs, during which time he co-chaired a global industry committee overseeing a developing standard for real-time securities trading, and then subsequently built an enterprise software company in that same space which he grew for 8 years before selling to NYSE.