• Intelligent software: programs with cognitive abilities, such as the software to drive autonomous cars and aircraft, or recommendations to persuade you to buy just one more pair of shoes
• High-performance software: from the largest computers to the smallest, smart engineers are squeezing the last drops of performance out of the hardware
• Virtual reality, augmented reality, mixed reality: from Hollywood to your smartphone your world has never looked so beautiful
• Big data: recent years have seen an explosion in the number of devices, some scarcely recognisable as computers, recording data about the material world, and a corresponding explosion in the storage and processing of the data
• Green computing: not only software systems to help scientists to understand how the planet is warming up, but software to reduce the demand for energy too, and even software designed to need less power to operate itself.

Looking at more specific examples:

1. Developing a New Algorithm or Process

This is a classic example of qualifying R&D. If your company is developing a new algorithm to solve a technical challenge that is not readily achievable by a competent professional in the field, this is likely to qualify. This includes:

• Designing and testing a new machine learning algorithm to improve the accuracy of predictive analytics.
• Creating a novel data processing method to handle big data more efficiently than current industry standards.
• Developing a new compression or encryption algorithm.

2. Overcoming Technical Challenges in Systems Integration

Simply integrating two off-the-shelf software products does not typically qualify. However, if the integration presents a significant scientific or technological uncertainty that requires a unique, innovative solution, it may be eligible.

• Integrating multiple systems in a way that requires the development of new middleware or protocols to ensure seamless data flow and functionality.
• Developing a new API to enable communication between disparate systems where no existing solution is available.

3. Improving Performance, Reliability, or Scalability

Routine improvements don’t qualify, but if the project involves a systematic process to overcome a significant technological challenge related to performance, reliability, or scalability, it may be eligible.

• Developing new software architecture (e.g., a microservices architecture) to allow a system to handle a massive increase in users without a degradation in performance.
• Conducting experimental development to optimise code for new hardware, such as a multi-core processor or a new GPU, where the outcome is uncertain.
• Overcoming bottlenecks in a database or system to achieve a level of performance that is beyond what a professional in the field could easily achieve.

4. Software for New and Emerging Technologies

Projects that involve new and unproven technologies are a strong area for R&D claims, as they are likely to involve technological uncertainty.

• Developing software for artificial intelligence (AI), machine learning (ML), or neural networks.
• Blockchain development for a new application.
• Creating firmware for a new Internet of Things (IoT) device or robotics.
• Developing software for virtual, augmented, or mixed reality applications.

5. Failed Projects

It’s important to note that a project does not have to be successful to qualify. As long as the work was undertaken to resolve a technological or scientific uncertainty and involved a systematic, investigative approach, the expenditure can still be claimed even if the project ultimately fails.

What is generally not considered qualifying R&D?

• Routine maintenance, bug fixing, or minor software updates.
• Simply using existing tools, languages, and methodologies in a standard development environment.
• Basic customisation or configuration of off-the-shelf software.
• Activities that are market-driven or commercial in nature, such as creating a simple e-commerce website with standard features.