TL;DR:
- In 2021, Oleg Klimov, Vlad Guber, and Oleg Kiyashko founded Refact.ai to enhance GenAI coding.
- Refact offers customization and control, targeting enterprise adoption.
- Developers acknowledge AI’s impact on coding, but employers express privacy concerns.
- Refact competes with GitHub Copilot and Amazon CodeWhisperer, offering offline capabilities.
- It uses permissively licensed code, avoiding liability risks.
- Refact secured $2 million in funding and is set to generate millions in annual revenue.
- The platform aims to execute code autonomously and self-test, ushering in a new era of AI coding.
Main AI News:
In 2021, Oleg Klimov, Vlad Guber, and Oleg Kiyashko embarked on a mission to create a cutting-edge platform, Refact.ai, aimed at enticing more enterprises to embrace GenAI for coding. Their goal? To provide users with unparalleled customization and control over the coding experience. With nearly a decade of collaboration under their belts, Klimov and Kiyashko, both software engineers with expertise in AI-based systems for image recognition and security, recognized the transformative potential of AI in engineering.
As the tech landscape evolves, most developers acknowledge the seismic shifts driven by AI. A recent HackerRank poll revealed that 82% of developers believe AI will redefine the future of coding and software development. Many are already embracing this change, as 63% of developers, according to VC firm HeavyBit’s 2023 survey, now incorporate GenAI in their coding tasks. However, the same enthusiasm is not universally shared among employers.
A separate survey targeting C-suite and IT professionals in enterprises found that 85% of them harbor concerns regarding GenAI’s privacy and security risks. Notably, companies like Apple, Samsung, Goldman Sachs, Walmart, and Verizon have imposed restrictions on internal GenAI tool usage due to apprehensions about data compromise.
Refact, akin to GitHub Copilot and Amazon CodeWhisperer, stands as a formidable GenAI coding assistant. It can address natural language queries about code, make code recommendations, and fine-tune its performance within a given codebase. Describing Refact, Klimov likens it to a “strong junior engineer” or an artificial co-worker that enhances team productivity under supervision.
What sets Refact apart from most competitors is its independence from an internet connection. Remarkably, it doesn’t even transmit basic telemetry data, safeguarding user privacy. Klimov emphasizes the company’s commitment to fortifying controls and processes pertaining to data sources and usage, security, and privacy, cognizant of the challenges faced by enterprises.
Refact’s platform relies on compact, code-generating models trained on permissively licensed code—a pivotal competitive advantage. Unlike tools trained on copyrighted or highly restricted code, Refact avoids the potential liability of regurgitating such code when prompted. Notably, GitHub and Amazon have introduced settings and policies aimed at addressing IP concerns around GenAI coding tools, but the impact remains uncertain.
“We used permissive license code to train our models because our customers demanded it,” Klimov explains.
Refact’s privacy and IP-conscious approach has garnered substantial support, including $2 million in funding from undisclosed investors. The platform, available in a cloud-hosted plan starting at $10 per seat per month, is on a path to generate substantial revenue, with projections of earning “a few million” annually by the summer.
This achievement is particularly noteworthy considering the struggles faced by vendors like GitHub, where tools like Copilot reportedly incurred substantial costs for cloud processing overhead—up to $80 per user per month for parent company Microsoft.
Looking ahead, the London-based Refact team, consisting of eight dedicated individuals, is focused on enhancing the platform’s capabilities. They aim to enable Refact to autonomously execute code, carry out “multi-step” plans, and self-test code. With ample internal funding and a team of talented individuals eager to contribute to the AI revolution, Refact continues to flourish as a hub for innovation, poised to leave a lasting impact on the industry.
Refact can run offline, on-premise or in a cloud-hosted managed setup. Source: Refact
Conclusion:
Refact’s innovative approach to GenAI coding offers a compelling solution for enterprises. With a focus on privacy, security, and code quality, it addresses key concerns in the market. The platform’s success highlights the growing importance of AI in coding, while also emphasizing the need for robust privacy and IP protection measures in GenAI tools.
