COMMENT
How AI-assisted code remediation is a breakthrough in embedded software development
By Steve Howard, Perforce Software E
mbedded World is always a beacon for the latest industry innovations, and not surprisingly, AI-related developments will likely proliferate again in 2026. One of the latest is AI-assisted code remediation, which will help software developers fix coding errors, security vulnerabilities and standards compliance issues more efficiently and effectively. Although only recently available, AI-assisted code remediation is expected to soon become a standard part of DevOps processes.
To understand why this approach is such a breakthrough, it helps to have some background. Embedded software developers have long used static analysis tools to automate much of the process of finding issues, and this approach has certainly been less time-consuming than manual code reviews. However, developers still spend 30-50 per cent of their debug time interpreting static analysis results and researching fixes.
Some developers have experimented with using AI-coding assistants to see if those could speed up the process and support better productivity, but this can lead to a trade-off in quality control. Carnegie Mellon recently presented content that indicated fix accuracy levels as low as just 20 per cent, when this is not done well. Clearly, 20 per cent accuracy is not sufficient, or even usable, especially in safety-critical environments such as medical, automotive, defence and aviation embedded systems.
More context means more accuracy
This is where AI-assisted remediation, when combined effectively with deep inter- procedural static analysis data, can make a significant difference. The static analysis results not only alert embedded software developers to hard-to-detect complex coding issues, but also feed this rich context- aware data to the AI Code Assistant, leading to more accurate fix recommendations. The AI proposed change is then presented as a code diff view, with ‘this is what would be removed’, and ‘this is what would be added’ or ‘this is what would be modified’ alongside
42
all the context and resolution details in the AI chat window.
State-of-the-art static analysis is designed to provide full data-flow and control- flow traces. It knows, for example, where a variable first occurred, how its value changed and which execution paths led to a fault such as a buffer overflow. Consequently, this intelligence enables the AI-assisted code remediation tool to make a more accurate and reasoned proposal for the fix. In this model, accuracy jumps to 72-90 per cent, and Perforce’s own lab research has found well above 90 per cent accuracy in real-world cases. So, in this way, AI can give embedded teams confidence that it is an asset, rather than being a hindrance that relies on an incomplete picture.
Optimal ‘shift left’ and MCP- compatible
In addition to delivering higher quality code, MCP-based AI assisted code remediation also enables productivity improvements for embedded developer teams shifting left. This is ultimately through the concept of find- and-fix early in the software development lifecycle, which has become an integral part of DevOps best practices.
Thanks to AI assisted code remediation, developers are spared additional workload, since they are provided with accurate solutions to new problems, rather than being presented with large volumes of
FEBRUARY 2026 | ELECTRONICS FOR ENGINEERS
overwhelming information. Plus, all of this happens within the developer’s integrated development environment (IDE) so there is no distracting context-switching. Also, as soon as a proposed change is made, the analysis reruns, incrementally, to verify that the issue has been fixed and that no new issues have been introduced. So, this creates a continuous, self-checking and much safer environment for creating higher- quality code that will not break other parts of the system.
However, crucially, it is still humans who remain in charge of what happens, through the developer-in-the-loop acceptance. While AI may be working its magic in the background, AI-remediation is fundamentally guided and controlled by software engineers, who have full visibility of what has happened, what will change and the reasoning behind AI’s recommendations. This gives embedded software teams the assurance that AI is a valuable assistant, rather than the ultimate decision-maker. With AI-assisted remediation, developers can not only analyse-as-they-code but also remediate-as-they-code, leading to more accurate, safer software, optimally shifting left defect detection and remediation and improving development velocity, which in turn enhances DevOps processes in embedded projects.
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50
