The Arbitrum Growth Circle is live and we’re looking for industry people to participate.
Working in fintech or financial services and interested to develop your onchain stack, please reach out.
If you’re a Web3 business at any stage looking to GTM with fintechs, do the same!
Moments like today remind me why our stubbornness on privacy in @usewisp_io matters.
I'm trying to get back on track with supplements, which means I need real research from AI. And real research means laying out the full picture, not just the easy stuff, but the mood swings, the anxiety, the stress, the physical and mental health things I normally keep guarded.
For the first time in a while, I actually talked my heart out. Didn't self censor. Didn't hold back the uncomfortable details. Not because I suddenly stopped caring about privacy, but because I finally trust the stack: wisp architecture, inference, and CVM setups (@PhalaNetwork).
There's something deeply relieving about being able to lay your health bare to an AI and not have that quiet voice in your head warning you that it might leak, train somewhere, or show up later.
That peace of mind is the product. Everything else is just chat.
The biggest company in agentic finance may not have the best autonomous agents. It may be the company that best understands how to earn enough trust from ordinary people that they'll actually let AI touch their money.
Yesterday was a real cracker of an event @devfolio around the theme of sovereign AI. @rashmivabbigeri got speakers from @SarvamAI (India's frontier models) and @PeoplePlusAI (integrating LLMs into workflow)
It started with the definition
sovereign AI = Data + Model + Inference
The 1st parameter is simply whether data from orgs is safe & whether they have control over the product long term. This is a dealbreaker for sectors like defence & important for heavily regulated ones like finance or healthcare
2. The 2nd part on the model is most useful for localization. For eg people plus ai used sarvam in their pilot with Gujarati cooperative dairy farmers from amul simply because it understood gujarati the best, not because it was Indian.
It's also unrealistic to expect every country to make its own frontier model, so the best we could hope for is;
a. Every country gather's it's own indigenous language data
b. A frontier model can then be trained on the local data & made useful for the local context
AI4Bharat is an example of such a program with $10 million allocated to collecting local speech data, since frontier labs would not find it profitable to do so (although they would use the data if it were open)
3. The final part on inference brought forth some unique stories in working with governments. For eg some departments required installation on premises & then airgapping the servers so they don't interact with the internet and leak any info; in other cases they had to optimize with the existing hardware of a government department since we couldn't expect them to acquire blackwells or H100s. In one case they even needed edge models able to use the phone as hardware since internet access isn't guaranteed (such as with rural health workers or soldiers on a mission)
Overall the vision of every country building its own frontier model felt a bit like the failed economic theory of import substitution. What was more realistic was every country building out its own corpus of speech data and then forking a frontier model to let it work in their language. I'm hopeful that sarvam can carve out a space for themselves in the developing world with this niche, and that people plus ai can go the whole mile and actually deliver the productivity benefits that AI is supposed to bring society at large
The Bulgarian government has signed this week a Memorandum of Understanding with the Bulgarian leading space company EnduroSat. In the Dobroslavtsi industrial zone, on the disused former military airfield, the company will build a large-scale production centre for satellites.🚀
Unless I’ve mistaken this is a first-of-kind regulated insurance policy covering crime to such an extent in DeFi.
It was a huge effort to get here with over 12 months of work from Teo Kithara-Bigot, Adam Wickens, Alex Krasnow and guest appearances from Christian Ogden-Davies.
We feel this is the most robust regulated product out there in the market for DeFi lending markets and we hope this can provide foothold for more cover to grow as we expand.
I’m fed up with the direction our country is going.
So I'm running for Congress in Virginia's new 7th district to put working families first, hold the powerful accountable, and fight to build a future we can actually be proud of.
Learn more on our site linked below:
I'm running for US Congress. Seriously.
Imagine common sense crypto policy and regulation, aligned with American dynamism.
Check out our list of new Crypto Policies we will support, linked below.
https://t.co/TOylxM5Hk3