Talview Patents Alvy: Agentic AI Revolutionizes Proctoring, Outsmarts Cheating
Talview, a company based in San Mateo, California, has been granted its inaugural U.S. patent for Alvy, its advanced Agentic AI proctoring solution. Awarded on July 15, 2025, as Patent No. US 12,361,115 B1, this patent marks a significant step in the evolution of online assessment integrity. Alvy is engineered to serve a dual purpose: rigorously detecting sophisticated cheating mechanisms in remote interviews and high-stakes online examinations, while simultaneously guiding candidates through a fair and transparent assessment process.
Co-invented by Talview CEO Sanjoe Tom Jose and CTO Subramanian Kailasam, the patented technology describes an AI-driven system that leverages large language models (LLMs) to monitor users during critical tasks like exams and interviews. This system analyzes real-time media data to identify anomalous behavior, trigger alerts, and, if specific thresholds are surpassed, terminate sessions to prevent fraudulent activities. Beyond its fraud prevention capabilities, Alvy is also designed to assist candidates by clarifying instructions, alleviating stress through empathetic AI interactions, and ensuring clarity at every stage of the assessment.
Sanjoe Jose, Talview’s CEO, highlighted the foundational difference of their approach, stating, “Most platforms are still patching legacy systems. Alvy reimagines proctoring by bringing genuine artificial intelligence that can think, adapt, and outsmart even the most sophisticated cheating tools. This patent validates that maintaining trust in the AI age requires intelligence, not just observation.” This perspective underscores Alvy’s departure from conventional proctoring methods.
Unlike traditional systems that typically rely on passive monitoring and rigid rule-based detections, Alvy operates as an agentic system. This means it is capable of perceiving, making autonomous decisions, and acting independently to uphold the integrity of remote processes. Its design specifically targets and counteracts advanced AI-assisted cheating tools, including platforms like Cluely, instances of ChatGPT misuse, and deceptive deepfake-driven impostors. Alvy’s core capabilities include autonomous decision-making, enabling real-time judgments without predefined rules; sophisticated deepfake and AI content detection; and advanced behavioral intelligence to recognize subtle patterns indicative of external AI assistance. Furthermore, its LLM-powered contextual understanding allows for nuanced threat assessment, and its adaptive learning mechanisms ensure it continuously evolves to counter emerging cheating methods. Critically, it also offers real-time, context-aware assistance to test-takers, enhancing their experience.
The impact of Alvy’s agentic approach is reflected in its reported performance metrics. The system is said to detect eight times more suspicious activities than legacy AI proctoring solutions. It reportedly recognizes 99% of events that are also flagged by human proctors, indicating a high degree of accuracy and alignment with human oversight. Moreover, the system contributes to a reported 35% increase in candidate satisfaction, attributed to its real-time and context-aware assistance, which fosters a more supportive assessment environment. This blend of robust security and candidate support positions Alvy as a significant advancement in the realm of online assessments.