Serial entrepreneur Pavlo Kharmanskyi launches Open Steps to help non-engineers build with AI coding agents
Startups Magazine Roshini Bains
Pavlo Kharmanskyi launched Open Steps, an open-source tool for people using AI coding agents. It translates the bots' jargon into plain English so non-engineers can tell what’s done and what still needs work.
Based on reporting by Startups Magazine, Roshini Bains — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Pavlo Kharmanskyi has spent more than two decades building companies, and his latest one is aimed at a problem he ran into himself: AI coding agents that can write code but still talk like engineers talking to engineers. His new project, Open Steps, is an open-source tool meant for founders, product managers and other non-technical builders who are increasingly leaning on AI to make software.
The idea is not to replace the coding agent. It is to make the output usable. Open Steps rewrites technical progress into plain language, points out what still needs doing, and flags the parts that need a human to step in. The project also includes tools that check whether work is actually complete, turn technical explanations into non-technical ones, identify next steps and verify claims made during coding sessions.
Kharmanskyi says he built it from direct frustration. In his own words, the agent would do the work and then report back in commit hashes and jargon, leaving him unable to tell whether the task was finished. That kind of gap matters more as AI tools move into the hands of people who do not have engineering backgrounds but still need to steer the process.
Open Steps is free under an MIT licence, and both its methodology and testing data are public. The project has already picked up more than 1,000 stars on GitHub. That is a modest number in internet terms, but it does suggest there is real interest in making AI development tools a little less opaque.
Kharmanskyi is not exactly new to shipping things people use. His first online community network grew to more than 10 million followers before a seven-figure acquisition, and he later put more than $1 million of his own money into FinSafe. His biggest current business, OnlyMonster, started in 2023 with one part-time developer and fewer than ten users in a private Telegram group, and now serves users in more than 100 countries.
My take — AI-written commentary, not fact-checked reporting
This is the sensible side of AI tooling: not more magic, just less nonsense. A lot of software now assumes the user is happy reading machine babble like a junior compiler whisperer, which is a great way to waste time and call it innovation. Open source is the right fit here because the problem is communication, and closed systems love hiding bad communication behind a slick demo.
Read more about this at: Startups Magazine