Recently, A US-based firm engaged Scintillation Research for a two-phase infringement analysis project:
- Phase 1: Patent mining to identify high-value U.S. patents suitable for enforcement.
- Phase 2: Preparation of litigation-ready claim charts.
Using a hybrid approach combining automation with expert analysis, Scintillation Research identified high-potential assets, mapped likely infringing products, and delivered confidence assessments for Evidence of Use (EoU) development.
Following our report, the client commissioned 20 litigation-ready claim charts. The Scintillation Research team executed the assignment using verified technical sources, defensible evidentiary materials, and a structured EoU methodology aligned with litigation standards.
When AI Became the Reviewer
To enhance quality control, the client deployed an AI-based review framework leveraging Gemini, ChatGPT, Claude, and Perplexity. The system evaluated:
- claim-to-evidence alignment,
- infringement logic,
- link authenticity, and
- completeness.
Initial AI-generated feedback suggested:
- low chart ratings,
- missing information, and
- absent reference links.
How the Scintillation Research team Responded
Rather than treating AI feedback as definitive, the team conducted a structured validation review.
Inconsistent AI Ratings
Cross-platform testing revealed significant variability in ratings. Repeated prompts produced different outputs, demonstrating that AI scoring alone cannot serve as a reliability benchmark.
Suggested Missing Links
Each link was verified. Several were inactive, corrupted, or unrelated. All valid references were already included in our submitted materials.
Flagged Information Gaps
The AI system failed to interpret information already present in the documentation. Scintillation Research highlighted and re-shared the relevant evidence for clarity.
Outcome
After reviewing Scintillation Research’s analysis and clarifications, the client confirmed the claim charts were accurate, complete, and litigation-ready, and formally approved the deliverables.
Market Insight
AI can accelerate review workflows — but expert validation remains indispensable.
This engagement reinforced that defensible IP enforcement work requires both advanced tools and seasoned domain expertise.
