Project description DEENESFRITPL Cost- and time-efficient AI tool to minimise coding errors in software programmes Coding errors are almost certain to occur because each software programme has millions of lines of code, and therefore a single programme requires innumerable code reviews and fixes. Today’s code review procedure is very costly, takes too much time and isn’t always successful in fixing the code. The EU-funded DC-IR project will deliver a platform that uses AI to automatically carry out software code reviews and provide suggestions based on how similar code-related issues were resolved in the past. The platform will also suggest code improvements for programmers and answer practically all software queries. Users are expected to save about 20 % in development time. Show the project objective Hide the project objective Objective Software has become very complex with time, with each software program having millions of lines of code. With so many code lines, it is close to impossible not to make errors when coding. The number of code defects rises proportionally with more code lines, (approx. 10-20 defects per 1,000 lines of code). This necessitates millions of code reviews and code fixes for a single software program. The current code review process is very expensive, time consuming (with companies like Google spending >25% of their time on code reviews) and often does not guarantee success in fixing the code. The software industry needs a cost & time-effective code analysis tool that is unlimited in detectable code errors and programming languages.Our solution is DeepCode AI Code Review (DC-IR), an Artificial Intelligence (AI) platform that automatically performs reviews on software code and provides suggestions based on Big Code learnings (how others solved similar code related problems). Our platform is trained from millions of Open Source repositories (billions of lines of code; thousands of frameworks & millions of code fixes) and uses these data sets to suggest code improvements for programmers. DC-IR integrates many levels of program code analysis into proprietary Machine Learning (ML) representations which are used by powerful ML techniques to create Data Sets that can answer almost any question about a software in a language independent manner. DC-IR offers a full set of services for code optimisation with solutions for code fixes & quality assurance. DC-IR enables developers to save 20% of development time, leading to savings of €11,856 annually per developer, which compounded globally can save the industry >€52Bn annually.DC-IR is at an advanced stage of development with a Beta version already deployed and having more than 5k users, some using paid licenses. During Ph1, we will develop a road map to finalise DC-IR and Ph2 will see us developing and validating the market versión. Fields of science social sciencessociologyindustrial relationsautomationnatural sciencescomputer and information sciencessoftwaresoftware developmentnatural sciencescomputer and information sciencesartificial intelligencemachine learning Programme(s) H2020-EU.2.3. - INDUSTRIAL LEADERSHIP - Innovation In SMEs Main Programme H2020-EU.3. - PRIORITY 'Societal challenges H2020-EU.2.1. - INDUSTRIAL LEADERSHIP - Leadership in enabling and industrial technologies Topic(s) EIC-SMEInst-2018-2020 - SME instrument Call for proposal H2020-EIC-SMEInst-2018-2020 See other projects for this call Sub call H2020-SMEInst-2018-2020-1 Funding Scheme SME-1 - SME instrument phase 1 Coordinator DEEPCODE AG Net EU contribution € 50 000,00 Address Kasinostrasse 10 8032 Zurich Switzerland See on map Region Schweiz/Suisse/Svizzera Zürich Zürich Activity type Private for-profit entities (excluding Higher or Secondary Education Establishments) Links Contact the organisation Opens in new window Participation in EU R&I programmes Opens in new window HORIZON collaboration network Opens in new window Other funding € 21 429,00
