Unlike other AI systems SES learns code of the current system and leverages this code to build new code, re-using existing code; this results in less code being developed and therefore is more efficient, produces more manageable code, minimises code duplications, reduces code churn and defects. With a feature rich toolset SES can identify code duplicates, dead code, auto produce documentation and suggest code fixes.
A report by
GitClear in 2023 found that prior to the Copilot launch only 3% of code was duplicated code, 1 year after the launch this had jumped to 9%. This poses huge problems in the manageability of code. GitClear also found after Copilot launched Code Churn (% of code reverted, quickly after launching; largely due to defects) increased by 39% on the previous year. Whereas SES AI does not experience the same issues, because the core fundamental of SES is to leverage existing pre-tested and reliable code.