Meet Philip
Hi, my name is Philip Lankes, and I’m the founder of Rookion.
Rookion basically combines three things that have been with me for a long time: chess, AI and building digital consumer products.
I first got into chess as a kid through my father. He is an experienced league player and still spends several hours a day on chess. Professionally, I later spent several years working on digital B2C products, most recently in a leadership role covering product strategy and product management.
Today I’m building Rookion as a solo founder. A lot of what goes into the product comes directly from conversations with players, coaches, clubs, schools and, of course, from my father’s decades of chess experience.
What inspired you to create Rookion?
It actually started as a birthday present for my father.
I didn’t want to give him another chess book or traditional chess program. I wanted to build something that could help him understand his own weaknesses and thinking patterns.
Chess engines are extremely good at telling you which move was better. What they often don’t tell you is why you personally didn’t find that move, what you misunderstood in the position or which thinking mistake led to your decision.
That was the gap I wanted to close.
While building the first version, I started using it myself after a longer break from chess and had quite a few aha moments. Then I tested it with beginners, improving players, club players and coaches and kept seeing the same problem: people play a lot of chess, but often don’t really understand why they keep repeating the same mistakes.
What started as a personal birthday present slowly turned into Rookion.
What makes Rookion fundamentally different from traditional chess-learning tools?
Most traditional tools are very good at answering what happened.
They show you the best move, an evaluation bar, a variation or a list of mistakes.
Rookion tries to understand why it happened.
It goes through your own game with you in a Socratic dialogue and asks what you were thinking in the critical moments. What were you trying to achieve? What did you expect your opponent to do? What did you overlook?
The coach then builds the explanation around your answer instead of simply showing you a generic engine line.
So instead of getting a static analysis report, you have a conversation that is much closer to working through the game with a coach.
Rookion is also deliberately not just an openended AI chatbot. It’s built specifically for coaching your own games, and it’s designed to stay true to what actually happened on the board.
How does the Socratic dialogue actually work after a player makes a mistake?
After a game, Rookion identifies the most important moments where the evaluation changed significantly.
For each one, it shows you what you played and what would have been better. Then, before explaining anything, it asks what you were thinking when you made the move.
You answer in your own words, and Rookion works from there.
Maybe your idea was actually good, but you missed one tactical detail. Maybe you completely misunderstood what your opponent was threatening. Or maybe you were focused on one side of the board and ignored something else.
The goal is to identify the mistake in the thinking process, not just the move itself.
You can then ask follow-up questions, challenge the explanation or explore alternatives. At the end, you get a summary of the important lessons from the game.
How important is playing actual games versus studying analysis for chess improvement?
I think you need both.
Playing is where your real decision-making happens. But simply playing hundreds of games without reflecting on them often means repeating the same mistakes again and again.
For me, the really valuable loop is:
play → reflect → understand → apply it in the next game
And your own games are probably one of the best sources of training material because they show exactly where you struggle rather than where another player struggles.
That’s also why Rookion starts with your own games – either games you play directly against Rookion or games you import from Chess.com, Lichess or PGN.
What are the biggest technical challenges in making an AI understand chess and communicate like a coach?
There are really two different problems.
The first one is chess correctness.
Large language models can talk about chess very convincingly while still being wrong about the actual board. They can misplace pieces, suggest illegal moves or explain variations that aren’t possible in the position.
Engines have the opposite problem: they are extremely precise, but an evaluation like “−2.3” doesn’t necessarily teach a beginner anything.
Building a coach that explains things like a human but stays true to the actual position was one of the hardest parts of building Rookion, and it’s where a lot of the development work went. Behind the scenes, Rookion has several safety nets and checks designed to catch wrong statements about your position before you ever see them. The details stay under the hood, but for the player the result is simple: the coach is built to talk about your real game, not an imagined one.
The second challenge is the coaching itself.
A good coach shouldn’t simply agree with everything you say or bury you in engine variations. The explanation also needs to fit the player.
A beginner might need to hear, “Your knight was undefended,” while a stronger player needs a completely different level of explanation.
Getting that balance right is something I’m continuously working on.
What happens when someone imports a game from Chess.com or Lichess?
You can import games from Chess.com or Lichess, import a PGN, take a photo of a paper scoresheet or play directly against Rookion. For Chess.com and Lichess, you simply enter your username and pick one of your recent games.
Rookion then analyzes the game, identifies the most relevant moments and starts the coaching process.
Instead of only receiving an engine evaluation, you work through those moments together with the coach and explain what you were thinking.
Afterwards, the game stays in your personal game collection together with the coaching history and game report, so you can come back to it later.
What features are you most excited to add to Rookion next?
One of the areas I’m most excited about is moving from analyzing single games toward understanding a player over time.
The long-term idea is that Rookion increasingly understands your recurring mistakes, strengths, weaknesses and patterns across many games.
From there, it becomes much more interesting: if Rookion recognizes that you repeatedly make the same type of mistake, it should eventually be able to give you targeted positions and training specifically for that weakness.
I also want to expand the training side around openings and strategic topics – not just showing players which moves belong to an opening, but helping them understand the ideas behind those moves and when they actually apply.
Ultimately, I want Rookion to become more of a personal companion throughout a player’s chess development rather than just something you open after one bad game.
What advice would you give founders building in the SaaS industry?
One of the biggest lessons for me has been that building the product is only one part of the job.
Today you can build and iterate on software faster than ever before. Getting the product in front of the right people, understanding why they stay or leave and finding repeatable distribution is often the much harder part.
I would also talk to users as early and as directly as possible.
Analytics can tell you what someone did. A real conversation often tells you why.
And especially with AI products, trust is incredibly important. One confident but completely wrong answer can damage the user’s trust much faster than ten good answers can build it.
So I’d rather ship something smaller that works reliably than add five features at once that I can’t properly control.
Did you enjoy our interview? Do you have anything to say to our community?
Absolutely – thanks for the questions and for giving me the opportunity to share a little more about what I’m building. 
And to everyone in the community: if you enjoy chess and feel like you keep making the same mistakes without really understanding why, feel free to try Rookion.
You can start for free, play a game or import one of your own games and go through it with the coach.
And most importantly: please tell me what you think.
Rookion is still a young product and I’m building it very closely together with the people actually using it. If something doesn’t work, you miss a feature, you find a bug or simply have an idea that would make Rookion more useful for you, message me anytime.
A lot of what exists in Rookion today came from exactly that kind of feedback.
