Probably
Works
How business works. And how it changes.
Two founders. Strong opinions. Honest footnotes.
*Results may vary
I Thought AI Would Free Me. I Was Wrong.
Oksana and Konstantin compare which tasks AI actually saves them time on, where generated code still needs engineering work, and how faster tools can lead to longer hours.
26 min · 7 chapters
Episode details and chaptersClaim. Test.
Verdict.
One of us says something bold. The other one pushes back. Then we decide what we actually believe, for now.
- Chapter 01 / 03 The claim *What are we actually assuming?
- Chapter 02 / 03 The test *What breaks first?
- Chapter 03 / 03 The verdict *Provisional. Obviously.
Episodes
-
EP.06 26 min I Thought AI Would Free Me. I Was Wrong.
Oksana and Konstantin compare which tasks AI actually saves them time on, where generated code still needs engineering work, and how faster tools can lead to longer hours.
-
EP.05 47 min CEO? More Like Chief Everything Officer
What happens when an introvert who wants to write software ends up running the whole company?
-
EP.04 29 min Are You Automating Things That Don't Matter?
Are you falling into the automation trap? Learn when to stop building complex systems and start focusing on real results.
-
EP.03 32 min Why AI Won't Replace Your Design Team (Yet)
If AI can produce average work, what should a human team bring beyond it?
-
EP.02 33 min The Reality of Building AI Products: It’s Not Just Prompts
A week's AI token allowance can disappear in a day. But the bigger cost may be the software—and information—we stop questioning.
-
EP.01 33 min AI vs. Humans: Who Should Control the Code?
You asked an AI agent to fix the padding. It changed the login instead. Who pays for the mistake—and why are you still paying for the tokens?
Articles
- Workflow automation ·
How to decide what to automate: a break-even checklist
Estimate the real payback of a workflow automation, account for exceptions and maintenance, and pick a small pilot before building the whole thing.
- Software engineering ·
A code review checklist for AI-generated changes
Review an AI-generated patch for changes outside its scope, verify tests against the original bug, and decide who owns the merge.
Two founders.
Two companies.
Oksana Kovalchuk and Konstantin Semenenko explore how businesses are built, how markets move, and how people make decisions when the rules keep changing. Real experiences, healthy disagreements, and a bit of sarcasm.
Konstantin Semenenko
Entrepreneur, software engineering expert and technical problem-solver. Turns ambitious ideas into software and AI products people actually use. Passionate about AI and software architecture.
FAQ
What is Probably Works about?
Probably Works is an English-language podcast about building and running businesses, leading teams, and how AI changes design and software. The hosts test a claim against their own experience and end with a provisional verdict.
Who hosts Probably Works?
Oksana Kovalchuk is the founder and CEO of ANODA, a digital product design agency. Konstantin “Kostya” Semenenko is the co-founder and CEO of Managed Code, the engineering company behind Prostir.
Where can I watch or follow Probably Works?
Full episodes, timestamped chapters, and Shorts are available on YouTube. The RSS feed lists full episodes with dates, summaries, and chapter links.
What is the difference between ANODA, Managed Code, and Prostir?
ANODA helps with product design and UX. Managed Code builds custom software and AI systems. Prostir is a platform for building and publishing your own AI agents.
Topics
- AI's impact on business
- Marketing & sales
- Finding the right customers
- Running a company
- Teams, ownership & seniority
- Tech bubbles & business history
Bring a claim
Got an opinion about business, AI or teams you want us to test? Leave it in the comments of the latest episode. The best ones make the show.
*Results may vary.




