On the Docket: 4 Ways to Operationalize the Process with AI
December 01, 2025 by Henry Sal
Only 5% of pilot artificial intelligence (AI) programs succeed and deliver meaningful results, according to a recent MIT report.
This statistic isn’t meant to alarm or deter courts. It’s a reminder for agencies and vendors alike that AI is meant to bring purpose to our work. Forget its flashiness or novelty. AI is intentionally designed to solve real problems for real people.
The report suggests that when integration with existing business isn’t thoughtfully considered, that’s when programs fail. It’s not the quality of the AI model that’s at fault; it’s the approach and defining measures of success that are overlooked.
Alternatively, successful AI programs excel because leaders concentrate on one pain point, understand the why behind a solvable problem, and execute strategically in line with desired outcomes.
Sophisticated court leaders operationalize AI with a heightened sense of practicality. A precise, problem-centric approach … this is the path to repeatable and confident justice.
Automation with Purpose
Consider the routine, error-prone, and time-consuming process for reviewing and docketing filings. This is an equitable problem for courts to solve. It is ripe for AI and automation, and court leaders across the country are seeing the value. They are automating their court document processing workflows to eliminate data entry errors, reduce the case backlog, fill the productivity gap, and free up staff for more valuable work.
After working with courts for decades, I’ve witnessed and helped shape several of these deployments first-hand. Operationalizing AI within existing court workflows to resolve case backlogs encompasses a measurable, and often scalable, program.
Featuring jurisdictions across the country, here are four realistic ways courts are implementing AI-enabled document processing:
1. High Volume, Low Complexity
A reasonable entry point for court document automation is low-hanging fruit: high-volume and low-complexity case filings. These are the filings that make up the mundane tasks of the backlog, much like affidavits, certificates of service, and responsive pleadings.
One of our first clients, Palm Beach County, Florida, focused on automating their high-volume, low-complexity filings so that clerks could focus on more meaningful work. As they saw success, they continued iterating on their automation program by introducing additional case types. They are currently automating more than 50,000 documents per month, and “human-in-the-loop” reviews are completed for about half of them.
2. High Volume, High Complexity
Going the opposite direction, a recent county we worked with concentrated on the noise — the high-volume and high-complexity documents that were clogging up the filing system. This county selected ten filing codes that were so time-consuming to review and docket that making a dent in the backlog seemed unfeasible. Automating these codes helped clear the bottlenecks in the filing queue and enabled clerks to complete simpler tasks. They are now projected to automate more than 10,000 envelopes per month.
We also typically categorize anything surrounding case initiation as high-volume, high-complexity. Starting the document and filing processes from scratch requires a highly skilled and well-trained AI agent. We have found it most successful to automate the high-volume, high-complexity filings later in the implementation process, as a phased approach allows agents to learn your workflow and produce better results.
3. Low Volume, High Complexity
Shifting gears slightly, low-volume and high-complexity filings are another focus area for courts, which typically include probate and specialty court filings. Though not as common, these filings are especially redundant and cumbersome — each taking a clerk around 10-15 minutes to process manually. Petitions for guardianship, estate inventories, or mental health petitions are frequent examples.
4. Low Volume, Low Complexity
And finally, we have low-volume and low-complexity filings that are repetitive but not overly complicated. For example, change of address notices, name change requests, or notices of appearance or withdrawals of counsel. These are simple and usually uncontested.
One county we worked with automated their low-volume and low-complexity filings to help clerks become comfortable. This rollout strategy — starting small and simple — resulted in an 85% accuracy rate. Not only did the AI program alleviate clerks’ administrative burden, but it also garnered a quick win by building staff trust. They’ve since moved on to higher-volume work after seeing the program’s accuracy.
Final Thoughts
Every court and the community they serve are unique. The most suited automation approach depends on how court leaders compare their current processing volume and metrics against their desired measures of success. And more importantly, successful AI programs aren’t born overnight. Automation is often best approached in phases, scaling programs over time by moving from volume to complexity. Those who succeed continuously measure and track their accomplishments to identify where and how else to build confidence in their case-processing workflow.
About the Author
Henry Sal
Senior Director of AI Automation
Henry Sal is the Senior Director of AI Automation at Tyler Technologies. With a software development career that spans over four decades, Henry is responsible for Tyler’s state-of-the-art artificial intelligence used to automate court document workflows. Henry is a thought leader in his field and has shared his insights on machine learning and document understanding through the various industry partners, including the National Center for State Courts (NCSC), the Association for Information and Image Management (AIIM), the Property Records Industry Association (PRIA), and the PDF Association.