In 2021, Oleg Klimov, Vlad Guber and Oleg Kiyasko set out to co-create a platform, Refact.ai, that could convince more companies to embrace GenAI for coding by affording users more customization \u2014 and control \u2014 over the experience.<\/p>\n
Klimov and Kiyasko had worked together for nearly a decade building AI-based systems for image recognition and security systems. Guber knew Kiyashko from childhood; they were neighbors in the South Ukranian town of Yuzhnoukrainsk.<\/p>\n
\u201cIt was clear that AI would change the very notion of what engineering is,\u201d Klimov told TechCrunch in an email interview. \u201cAs software engineers at heart, we decided we need to be in the best position to live through it \u2014 creating an independent system for software engineering.\u201d<\/p>\n
Most devs acknowledge the AI-driven seismic shifts occurring in their profession. Eighty-two percent responding to a recent HackerRank poll<\/a> said that they believe AI will \u201credefine\u201d the future of coding and software development.<\/p>\n Refact taps generative AI models trained on permissively licensed code to power its code-suggesting platform for devs.<\/p>\n<\/div>\n The majority are embracing the change, with 63% of devs in VC firm HeavyBit\u2019s 2023 survey<\/a> saying that they\u2019re now using GenAI in coding tasks. But employers are more skeptical. In a separate survey<\/a> of enterprise C-suite and IT professionals, 85% expressed concerns about GenAI\u2019s privacy and security risks.<\/p>\n Companies including Apple, Samsung, Goldman Sachs, Walmart and Verizon have gone so far as to limit<\/a> internal use of GenAI tools over fears of data compromise.<\/p>\n So how\u2019s Refact different? It runs<\/a> on-premise, Klimov says.<\/p>\n Like GitHub Copilot, Amazon CodeWhisperer and other major GenAI coding assistants, Refact can answer natural language questions about code (e.g. \u201cWhen was this dependency last updated?\u201d), recommend lines of code and fine-tune to improve its performance with a given codebase.<\/p>\n \u201cOne way to think about it is as a \u2018strong junior engineer,’\u201d Klimov said, \u201cor an artificial co-worker on a team that\u2019s productive but needs supervision.\u201d<\/p>\n However, unlike many \u2014 if not most \u2014 of its competitors, Refact doesn\u2019t need an internet connection. It doesn\u2019t even upload basic telemetry data, Klimov claims.<\/p>\n Refact can run offline, on-premise, or in a cloud-hosted managed setup.<\/p>\n<\/div>\n \u201cWe\u2019re developing better controls and processes around sources and uses of data, security and privacy as we\u2019re aware of the challenges that [enterprises] face and want to ensure the integrity of our customers\u2019 information and innovative breakthroughs,\u201d Klimov said.<\/p>\n Powering Refact\u2019s platform are compact, code-generating models trained on permissively-licensed code \u2014 another key competitive advantage, Klimov claims. Some code-generating tools trained using copyrighted or otherwise restrively-licensed code have been shown to regurgitate that code when prompted in a certain way, posing a potential liability risk for the companies deploying them (at least according<\/a> to some IP experts).<\/p>\n Vendors like GitHub and Amazon have introduced settings and policies aimed at allaying the fears of companies wary of the IP challenges around their GenAI coding tools. But it\u2019s unclear that they\u2019ve made much headway. In a 2023 survey<\/a> of Fortune 500 businesses by Acrolinx, nearly a third said that intellectual property was their biggest concern about the use of generative AI.<\/p>\n \u201cWe used permissive licence code to train [our models] because our customers demanded it,\u201d Klimov said.<\/p>\n Refact\u2019s privacy- and IP-conscious approach helped it to raise $2 million in funding from undisclosed investors \u2014 and net ~20 pilot projects with enterprise customers. Klimov claims that the platform, which is also available in a cloud-hosted plan that starts at $10 per seat per month, is revenue-generating and currently on track to earn \u201ca few million\u201d annually by this summer.<\/p>\n That\u2019s impressive, considering vendors like GitHub have struggled to make a profit on their code-generating tools. Copilot was reportedly<\/a> costing GitHub parent Microsoft up to $80 per user per month as a result of the associated cloud processing overhead.<\/p>\n Refact is among the growing cohort of GenAI tools for programming.<\/p>\n<\/div>\n The focus for the London-based, eight-person Refact team in the near future is upgrading Refact to run code autonomously, execute \u201cmulti-step\u201d plans and self-test code.<\/p>\n \u201cWe\u2019re actively working on a next-generation AI assistant \u2013 one that\u2019ll debug the code it writes and operate on any large codebase,\u201d Klimov said. \u201cWe\u2019re well-funded internally and have the necessary capital to continue building the product \u2026 We\u2019ve never benefited from an abundance of funding or from the venture capital frenzy that took place in previous years, but what\u2019s really benefited us is the availability and eagerness of very talented people who are looking to join the AI revolution \u2014 and who saw in Refact as a place to thrive and develop something that can have a lasting impact.\u201d<\/p>\n<\/p><\/div>\n![\"Refact\"](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2024\/01\/Refact-2.png\")
<\/p>\n![\"Refact\"](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2024\/01\/Refact.png\")
<\/p>\n