X23.ai: Governance aggregator + assistant

[daveytea]

GM!

I’ve spoken to a number of people in the Arbitrum community who have been using our product x23.ai to keep up to date with governance. I’d love to get more feedback from users and the community on how we can make the product even better for Arbitrum DAO and delegates.

One of the features we’ll complete soon are synthetic delegate profiles, so the community can see how and why delegates vote certain ways. Another upcoming feature will be proposal evolution, so readers can easily follow the lifecycle of proposals, from discussions to onchain vote.

I’d love to hear any other Arbitrum specific challenges that we could try to crack. E.g. in the previous STIP applications, there were so many that it was difficult to process all of them consistently. Another could be that developers may be confused as to where they should start… stylus, orbit, nova, one..? What would you like to see solved?

---

As a quick overview for those that don’t know what x23.ai does, we:

• Aggregate forum, snapshot, github, and official data
• Use AI and LLMs to condense and bring more understanding to the data

The results are a newsfeed, summarised articles, instant search, and a chat bot.

I made a video to showcase these features for the community here: https://youtu.be/BsGXSn-tkYU

The custom Arbitrum chatbot we just soft launched on Telegram here and i think it works rather well. For example, ask it about the constitution.

The aggregated Arbitrum newsfeed can be found here.

---

P.S. because we aggregate data across 29 different DAOs, we’ve found that Arbitrum is by far the most active DAO community

[daveytea]

GM! Just an update that we’re now covering onchain proposals, so you can easily see all current and historic proposals, both offchain and onchain, here: Uncover Arbitrum Voting

[daveytea]

GM! We just released a feature that is hopefully helpful for a lot of people: Timelines.

It helps people better understand the context of discussions and proposals.

For example, you can see how the Watchdog discussion has evolved into an onchain proposal:

Or you track how the MSS program has evolved and ultimately closed down:

I also posted about this on X if you want more examples and context.

As always, i’d love to hear any and all feedback!

[paulofonseca]

@daveytea

and ultimately closed down

the MSS didn’t closed down yet @daveytea… =) the proposal to close it down was only posted 5 days ago in the forum, and didn’t even went up for a vote yet. Wind Down the MSS + Transfer Payment Responsibilities to the Arbitrum Foundation

this is one of the dangers of AI summarized governance btw… people will infere too many conclusions, too early, that are not true, and then act like if it was true. that’s dangerous. Delegates should use way more critical thinking when ingesting AI summarized governance feeds.

[cp0x]

@paulofonseca

this is one of the dangers of AI summarized governance btw… people will infere too many conclusions, too early, that are not true, and then act like if it was true. that’s dangerous. Delegates should use way more critical thinking when ingesting AI summarized governance feeds

I agree, I recently encountered this problem myself

I asked AI for information, he answered me in detail, and then when talking to people it turned out that AI had clearly messed up something with the information and the numbers he provided were fully incorrect.

Now every time I ask AI for something, I have to double-check the information, because there is no guarantee that this is correct data

[daveytea]

Thanks for the comments guys, you bring up an important point.

@paulofonseca In this case it is actually the human misunderstanding (me!) as I just browsed over the headline of ‘Winding down’, however in the AI summarised article you’ll see that it makes clear that it hasn’t passed yet. This was just used as an example to show the evolution of the discussion and ultimately the proposal step, but alas you’re correct in that I should have been more nuanced in explaining this particular proposal evolution.

@cp0x This can happen when using generalised AI solutions, hence why we’re building crypto specific models where it is able to answer in detail, accurately. It is ofc still a work in progress, but i’d love you to give x23.ai Arbitrum Wizard a try: Telegram: Launch @ArbitrumWizardBot and see if you can stump it.

[paulofonseca]

@daveytea

@paulofonseca In this case it is actually the human misunderstanding (me!) as I just browsed over the headline of ‘Winding down’, however in the AI summarised article you’ll see that it makes clear that it hasn’t passed yet. This was just used as an example to show the evolution of the discussion and ultimately the proposal step, but alas you’re correct in that I should have been more nuanced in explaining this particular proposal evolution.

yeah I know! that was my point above… humans reading just the headline and running with it, is a problem that precedes this AI age =)

but AI tools usually augment the risk of that type human error, instead of trying to mitigate it.