QUALCOMM Incorporated has secured a major milestone in the Artificial Intelligence, Software, Cryptocurrency and the Cloud industry with a newly patented execution plan technology. This innovation focuses on a recently awarded patent titled ‘Executing queries in computing systems using execution plans generated by generative artificial intelligence models’. The patent describes a machine learning technique designed to refine execution plans and dynamically resolve computing errors by adjusting action granularity.

Outstanding Invention: Patent of the Month

Because it represents an outstanding invention in its field, QUALCOMM Incorporated has been awarded Swanson Reed’s Patent of the Month for January 2026 in the Artificial Intelligence, Software, Cryptocurrency, and Cloud industry.

Patent Abstract

Certain aspects provide techniques and apparatus for executing queries in a computing system using machine learning models. An example method generally includes receiving a plan to satisfy a request in the computing system and event log data associated with execution of the plan. The plan generally specifies a first plurality of actions to be performed by the computing system at a first level of granularity. Using a plan refinement machine learning model, a refined plan is generated when the event log data indicates that execution of the generated plan results in one or more execution errors and the one or more execution errors are solvable. Generally, the refined plan specifies a second plurality of actions to be performed by the computing system at a second level of granularity, the second level of granularity being finer than the first level of granularity.

Meeting the USA R&D Tax Credit Rules

The development of the technology outlined in this patent is a prime example of activities that satisfy the IRS Four-Part Test for the United States Research and Development (R&D) Tax Credit:

• Permitted Purpose: The research was undertaken to create a new or improved function, specifically enhancing the reliability, performance, and error-resolution capabilities of computing queries via generative AI models.
• Technological in Nature: The foundation of the research relies heavily on the hard sciences—specifically computer science, artificial intelligence, and software engineering.
• Elimination of Uncertainty: At the outset, Qualcomm faced technical uncertainty regarding how an AI system could accurately identify the root cause of an execution error from event log data and autonomously adjust the granularity of system actions to bypass the error without human intervention.
• Process of Experimentation: The engineers had to undergo a systematic process of trial and error. This included designing the “plan refinement machine learning model,” training it on various event logs, testing different algorithmic approaches to granularity adjustment, and evaluating the success rates of the refined execution plans.

3 Practical Applications Qualifying for the R&D Tax Credit

For software and tech companies, applying the concepts of this patent in the real world involves significant development work. The following three practical applications would likely qualify as eligible R&D activities:

1. Autonomous Database Query Optimization: A software company developing a proprietary, AI-driven database system could use these concepts to build databases that auto-correct failed SQL or NoSQL queries. The iterative coding, testing, and algorithmic refinement required to train an ML model to dynamically break down failed data queries into finer, successful sub-queries would constitute qualified research activities (QRAs).
2. Self-Healing Cloud Infrastructure Automation: In cloud computing environments, developers building automated resource provisioning architectures can apply this method. If a high-level server deployment plan fails due to configuration errors, an AI model dynamically breaks the provisioning actions into smaller, finer steps (e.g., allocating specific microservices independently). The experimentation required to integrate and train this ML model within a custom cloud environment directly qualifies for the credit.
3. Smart Contract Execution in Cryptocurrency: Blockchain engineers could adapt this ML-driven plan refinement to optimize complex smart contract execution across decentralized networks. Resolving technical uncertainties around how to autonomously handle transaction failures, and developing the machine learning models to adjust the execution path and granularity of the cryptographic requests, would involve highly specialized experimentation eligible for R&D tax incentives.