TMAE Summit 2026License to Skill: Building Smarter Trademark Workflows with AI
Alt Legal Team | October 25, 2026
At the TMAE Summit on October 19-21, 2026, Kathleen Jose of Uber presented the session, “License to Skill: Building Smarter Trademark Workflows with AI.” Learn about additional sessions here. Here are some key takeaways from the session:
1. Think beyond AI prompts and start building reusable workflows. Rather than repeatedly asking AI to perform the same tasks, trademark professionals can create reusable AI “skills” that follow predefined instructions, processes, and guardrails. This approach can help teams save time, standardize work, and improve consistency.
2. Start small and build with a specific purpose. Not every task needs a sophisticated AI agent. Breaking complex processes into smaller, focused skills makes them easier to test, maintain, and reuse. These individual skills can then work together to support more complex trademark workflows.
3. Prioritize tasks that are repetitive, teachable, and low-risk. The best candidates for AI skills are tasks performed frequently, with clearly defined steps and outputs that can be reviewed before consequential decisions are made. Examples include extracting trademark watch notice information, preparing matter summaries, and checking filing packets for missing information.
4. Use six building blocks to create more reliable AI skills. Effective AI workflows should have a clearly defined purpose, context, workflow, resources, guardrails, and output requirements. Establishing these elements upfront helps ensure AI understands what it’s supposed to do, what information it should rely on, and where its responsibilities end.
5. Keep AI focused on organizing facts, not making legal decisions. AI can assist with gathering information, researching trademark use, identifying factual similarities and differences, and preparing materials for attorney review. However, legal conclusions, risk assessments, and recommendations about whether to oppose a mark should remain with qualified professionals.
6. Test AI workflows for failure, not just success. A workflow that performs well under ideal conditions may struggle with incomplete information, changing facts, or requests that push beyond its intended scope. Testing challenging scenarios can reveal weaknesses in instructions and guardrails, allowing teams to refine their skills before relying on them.
7. Human oversight remains essential. Even well-designed AI skills can make mistakes or produce unsupported conclusions. Trademark professionals should require traceable sources, explicitly identify unknowns, keep reference materials current, and review AI-generated work before taking consequential action.
Bonus Notes:
- Be very descriptive in how you title your skill so that others on your team know exactly what the skill does. Don’t choose a fanciful name like “Margaret!”
- AI has “golden retriever energy” especially ChatGPT – its default is to be eager to help and agree. In developing prompts, you can tell it to be critical.
Speaker
Kathleen Jose
, Senior Program Manager, Uber
Kathleen Jose is the Senior Program Manager of the Brand Protection Team at Uber. In this role, she oversees operational strategy, online brand protection, and domain portfolio management, while leveraging AI to streamline and enhance workflows. Kathleen has been a professional in the intellectual property field since 2015, including six years at Block, Inc. (formerly Square) supporting their global IP portfolio. She also currently serves as a Co-chair of the Events Subcommittee on INTA’s Trademark Administrators Committee. Kathleen is based in San Francisco, and in her free time enjoys baking, exploring new music, yoga and weight training.