In 2021, Oleg Klimov, Vlad Guber, and Oleg Kiyasko set out to co-create a platform, Refact.ai, that could convince more companies to adopt GenAI for coding, giving users more customization — and control — of the experience .
Klimov and Kiyasko had worked together for nearly a decade creating AI-based systems for image recognition and security systems. Guber knew Kiyashko from childhood. they were neighbors in the southern Ukrainian city of Yuzhnoukrainsk.
“It was clear that AI would change the very concept of what engineering is,” Klimov told TechCrunch in an email interview. “As software engineers at heart, we decided that we should be in the best position to live it – by creating an independent system for software engineering.”
Most developers recognize the seismic AI-driven changes happening in their profession. Eighty-two percent respond to a recent HackerRank voting said they believe artificial intelligence will “redefine” the future of coding and software development.
Refact uses generative AI models trained on permissioned code to power its code hinting platform for developers.
Majority embraces change, with 63% of developers at VC firm HeavyBit 2023 overview saying they are now using GenAI in coding jobs. But employers are more skeptical. In separate overview of business C-suite and IT professionals, 85% expressed concerns about the privacy and security risks of GenAI.
Companies such as Apple, Samsung, Goldman Sachs, Walmart and Verizon have gone so far as to restrict internal use of GenAI tools due to data breach fears.
So how is Refact different? It works on-premise, says Klimov.
Like GitHub Copilot, Amazon CodeWhisperer, and other major GenAI coding assistants, Refact can answer natural language questions about code (e.g., “When was this dependency last updated?”), suggest lines of code, and to improve its performance with a given code base.
“One way to think of it is as a ‘powerful junior engineer,'” Klimoff said, “or an artificial partner on a team that is productive but needs supervision.”
However, unlike many — if not most — of its competitors, Refact does not require an internet connection. It doesn’t even upload basic telemetry data, Klimov claims.
Refact can be run offline, on-premise or in a managed facility hosted in the cloud.
“We are developing better controls and processes around data sources and uses, security and privacy as we recognize the challenges that [enterprises] we face and want to ensure the integrity of our customers’ information and innovative discoveries,” said Klimov.
Powering Refact’s platform is solid code-generating models trained on permissive code — another key competitive advantage, Klimov claims. Some code generation tools trained using copyrighted or otherwise licensed code have been shown to return that code when specifically requested, posing a potential liability risk for the companies using them (at least according to some IP experts).
Vendors such as GitHub and Amazon have introduced regulations and policies aimed at assuaging the fears of companies wary of IP challenges surrounding GenAI coding tools. But it is not clear that they have made much progress. In 2023 overview of Fortune 500 businesses by Acrolinx, nearly a third said intellectual property was their biggest concern about using genetic AI.
“We used the authorized license code to train [our models] because our customers asked for it,” Klimoff said.
Refact’s approach to privacy and IP helped it raise $2 million in funding from undisclosed investors — and complete ~20 pilots with enterprise customers. Klimov claims the platform, which is also available in a cloud-hosted plan starting at $10 per seat per month, is generating revenue and is currently on track to make “a few million” annually by this summer.
That’s impressive, given that vendors like GitHub have struggled to make a profit from their code generation tools. He was a copilot According to reports costs GitHub parent Microsoft up to $80 per user per month as a result of associated cloud processing overheads.
Refact is among the growing cohort of GenAI tools for programming.
The focus for the eight-person London-based Refact team in the near future is upgrading Refact to run code autonomously, run “multi-step” designs, and self-audit code.
“We are actively working on a next-generation AI assistant – one that will debug the code it writes and work on any large code base,” Klimov said. “We are well funded internally and have the necessary capital to continue building the product… We never benefited from the abundance of funding or the venture capital frenzy that occurred in previous years, but what has really benefited us is the availability and willingness of very talented people who want to be part of the AI revolution — and who saw Refact as a place to thrive and develop something that can have a lasting impact.”
