This is a submission for the Notion MCP Challenge
When I started building AI tools for my wife, it was because she had outgrown Notion.
She uses...
For further actions, you may consider blocking this person and/or reporting abuse
Hey, nice one!
Just worried about accuracy on resyncing, but the rest looks cool.
I also signed up on Synapse, and noticed there is no payment plan. If that's intended or not, I don't know. Just wanted to let you know, to save you from big bills if this post blows up :)
One more thing. Maybe you also want to post this submission here:
Drop Your Challenge Submission Here
Nikoloz Turazashvili (@axrisi) ・ Mar 17
Thank you for your feedback. The project does not have plans or a payment method yet. It is still a tech demo and a personal tool (for my wife, it is an actual product, and she uses it daily).
But you are right, I will need to add some guardrails and finish the product in case someone else finds it useful, and as you mentioned, save me from big bills
Yeah, maybe limit usage or lock registrations for now :) don't lock me out though. I started my Japanese coaching. 😁😁 Promise won't abuse it too much
Thanks to your warning, I prioritized adding a simple internal plan system to prevent abuses and have a limit to let people play with that. Also, I added a demo account with a demo graph and conversations. Thanks again for the feedback.
I also checked vexrail, it looks like a really cool project, thinking of my side projects where I havent projecthem as a business, more like a personal product, and something I wanted to be live. The cost of AI forces me to think about a monetization strategy, mostly to make the project sustainable, and this idea of agentic ads makes total sense for this use case
I will keep an eye on it, exited what you are building there
Really appreciate that, and glad the feedback helped.
And yes, the monetization side of AI products is becoming impossible to ignore. A lot of “just for fun” projects eventually run into the same sustainability wall, which is exactly why I find this space so interesting.
Feel free to reach out at niko@vexrail.com. It would be great to have a friendly chat!
The insight isn't the technical integration; it's that you recognized the graph visualizer was for you, not for her. Most builders would have kept polishing it, assuming she just needed to understand it better. You watched her ignore it and rebuilt the interface in the place she already trusted. That's not feature development. That's listening.
You absolutely got it, AI integration and engineering details are great and fun, but for me, the most rewarding thing of buidling side projects or any project is seeing your ideas being pressure tested and figuring out what is actually useful and what's not.
I love this new reality where the cost of taking an idea and being able to experience it in real life is that low, and the tools of Notion MCP make this kind of integration of ideas faster and easier to implement.
Interesting perspective. I've had similar experiences in my projects.
Learned something new today. Thanks for putting this together!
Your wife can be happy to have you, it's a wonderful and inspiring project.
Thank you, I am happy to have her to be honest, she is the most demanding and picky user I have known, she never cut back saying something I build is useless but at the same time she will use it hardly when it is actually usefull so my hack to build thing that matter is build for her and I know I will build something interesting
The bidirectional sync solving the "raw graph is overwhelming" problem is sharp — we hit the same wall with knowledge graphs and ended up needing a structured view layer on top too.
I initially thought a graph database would be visually intuitive on its own, but we really need a presentation layer on top to make sense of that 'black box' for the user. I even considered structuring the storage to match the UI directly, but a graph is ultimately the best choice for long-term maintenance. It allows us to programmatically update nodes and their relationships as new data comes in. Relying strictly on semantic relationships without a proper graph structure would be a maintenance nightmare, so a two-layer approach—graph storage plus a UI layer—makes perfect sense.
Btw, I have seen your latest posts, you are on fire with your publication frequency, it seems you are having so much fun in this AI rabbit hole
Good experience.