Tech services firm Persistent is rolling out an AI coding companion to 16,000 engineers. Its CTO recognises the risk
“The legal structure of contracts between a services company and customers simply doesn't allow us to unilaterally activate this kind of tooling. We have to have customer buy-in."
Persistent Systems, an Indian technology services company, is making Amazon’s AI coding companion CodeWhisperer available to its 16,000 engineers – and will now have to persuade its clients across heavily regulated industries like banking and healthcare that it is safe to use.
CodeWhisperer is comparatively new to market: It was made generally available (GA) in April 2023. Like Microsoft’s CoPilot, it relies on generative AI to auto-complete code – layering in security scans, and the ability to attribute snippets of code that it generates back to their original source/open source project repository for IP/licence review.
Despite such critical guardrails and the potential for AI to help drive significant improvements in developer productivity, it is early days for adoption and resistance to letting AI generate code that may be used production is rife – for a smorgasbord of concerns that can be loosely clustered around code bloat, hallucinations, compliance, and legal risk.*
Persistent Systems CTO: The rewards are huge
Speaking at AWS Summit in London, Persistent Systems’ CTO Dr Pandurang Kamat told The Stack that there are “a lot of questions” from clients, including “‘what happens to my data?’” not least, he added, “because the first exposure to [generative AI] came from consumer-facing applications, where there was a Wild West around data confidentiality.
