Hey [@🏔 HELIKON 🏔],

The MVP is now fully open-access, you can test the proof-of-concept here and we’d be glad to hear your feedback.

This updated proposal is structured to fully implement open-source software, and we’ve already reduced the budget compared to the previous ref. 1679 submission.
Out of curiosity, do you have references to comparable AI tool development budgets that you’re basing your argument on? That would help us better understand your perspective.

Hey @The Ionian ,

Thank you for sharing your perspective. We’d like to clarify that the current vote for this ref 1717 is for the** beta development ** and in the roadmap we’ve made it clear that the beta will be built with open-source tech.

The MVP, on the other hand, was created as a proof of concept to test whether a consolidated dataset and AI could enhance governance user flows. At that stage, open source was not a strict requirement under the W3F DF grants guidelines.

That said, we’ve heard both your concerns and the broader community’s feedback and we are committed to developing the beta version of Clarys.AI with an open source–first approach going forward.

Hey @EzioRed and @OpenSquare/Yongfeng ,

Thank you for sharing your concerns, I am happy to clarify it for you in detail:
I believe it’s clear by now that AI tools (which many of us already use daily) can significantly increase productivity across almost any use case.

For Polkadot governance, developing a customized AI is not about replacing processes but about optimizing existing manual workflows. In the MVP, we focused on three user stories I personally encountered as a bounty curator:

1. Comparing proposals between bounty submissions and treasury submissions
2. Running quick accountability checks
3. Sourcing off-chain bounty submissions into one UI and comparing them with on-chain treasury submissions

The potential impact is tangible: using Clarys.AI could increase DV and curator efficiency by up to 300%, save over 120,000 hours of work and reduce costs by an estimated $6M in treasury work/time budget savings.

We’d be genuinely interested to hear more on your perspective. What makes you still feel that AI may not bring meaningful optimization to current manual workflows?