CEO @AgentCloakAI Sovereign AI with @InCountry Sovereign Data | @Layer3Press AI news agents with Nostr censorship resistance | 6x exits | Former CTO/CIO @CBSi
Miami Beach
Joined May 2008
- Tweets6.1K
- Following2K
- Followers8K
- Likes98.2K
Pinned Tweet
Introducing AgentCloak: Use any AI without sharing your real data.
Chinese AI services, ChatGPT, Claude, doesn't matter.
You probably try to hide details before asking: different names, fake numbers, no address.
But then the answer's useless because the AI is missing actual context.
AgentCloak runs in your browser.
It swaps your sensitive info for realistic fakes before sending anything, then swaps your real info back into the response.
You get what you need.
The AI gets nothing about you.
Works entirely in-browser.
Already trusted by some of the biggest companies in the world.
Now free to use.
agentcloak.ai/
I love this, back to the future!
TIL: Gen Z loves Yahoo ?!?
“Over 50% of our audience is Gen Z & Millennial. 75% of Gen Z is using Yahoo or one of our products.”
Jim Lanzone (@jlanzone) reveals how Yahoo’s legacy brand was able to capture the Gen Z audience:
“College Fantasy Football, we kinda broke the internet when we announced it this summer, and it has been amazing.. We had our 3 biggest days ever this year for NFL, but College—40% of those users are brand-new to Yahoo.”
“We have seen a 31% absolute point lift in brand love with Gen Z. That’s from our Cardi B campaign, from our Kyle Cooke, the Weatherboy campaign, from Summer House.”
“It has been incredible to see that Yahoo has a lot of latent brand love in there for it, & especially with younger audiences.”
Next up, replace MCP calls with API calls
We built a new harness using @typesafeai's Jev that cuts the cost of repetitive work by 90%. The harness learns the job as it runs, moving steps from LLM calls to code.
Running 100,000 compliance alerts costs >$290K on Opus 5.
With agentrun() we got it down to <$26K.
Devil is always in the details.
Majority of major holders vote to relinquish pro-ratas in order to fit a new investor = fine.
Choosing which of your existing investors get to participate and getting upset when other investors insist on also participating = not fine.
Huge for Miami tech to have CMU a top tech university set up shop, and especially the top robotics university!
More on #MiamiTech’s big news of the year. Carnegie Mellon to open a major Wynwood campus. We’ve updated this with more details! @refreshmiami #moamitech refreshmiami.com/news/its-a-…
This is pretty cool!
Here’s a real example.
I asked SocialCrank:
“Find me the founders in Miami who follow both @goblintown and @ajt so I can invite them to a dinner at Art Basel.”
113,034 @goblintown followers
→ 317 in Miami
→ 42 founders
→ 26 overlap with @ajt followers
All in a few seconds.
Maybe you can first get a yoghurt plant approved, and then switch it to a data center!
@Chobani is the #1 yogurt brand in the country. They could have chosen to invest anywhere — but they chose Pennsylvania.
This is just the latest proof that our economic development strategy is working.
Bringing Chobani to Pennsylvania will strengthen our dairy industry, create more opportunity for our farmers, and reinforce Pennsylvania’s role as a national leader in agriculture and manufacturing. That’s a win for everyone across the board.
It's a massive personal data pipe into AI
🤮 EU Member States propose a #ReverseRobinHood:
Take personal data from Millions of people and give it to the US-Billionaires. Data gathering for "AI processing" made automatically legal under the #GDPR.
#DigitalLandGrab
irishtimes.com/business/2026…
Building AgentCloak Desktop to protect your data from AI was fun.
I have to say making this video was even more fun!
Introducing AgentCloak: Use any AI without sharing your real data.
Chinese AI services, ChatGPT, Claude, doesn't matter.
You probably try to hide details before asking: different names, fake numbers, no address.
But then the answer's useless because the AI is missing actual context.
AgentCloak runs in your browser.
It swaps your sensitive info for realistic fakes before sending anything, then swaps your real info back into the response.
You get what you need.
The AI gets nothing about you.
Works entirely in-browser.
Already trusted by some of the biggest companies in the world.
Now free to use.
agentcloak.ai/
AI safety isn't that complicated. There needs to be a set of transparent controls a model should pass prior to ship. Could be government like NIST Cyber or from industry like ISO 27K. This really isn't complicated. It doesn't slow down development, but it does slow a release by adding an evaluation step and remediation. Like all software.
One interesting potential change is human review. Currently a secure SDLC includes human review of all changes. Perhaps we can soon trust an AI to perform the review. But a model should not review itself! :)
I'm all in on cyber, biological, etc safeguards.
We need a very clear definition of what "alignment" means. AI as a pursuasive tool when there is only a single perspective is terrifying.
Elon joining in should maintain the only politically balanced AI. At least in the US, for the time being.
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: darioamodei.com/post/we-must…
Peter Yared retweeted
State social media monitors viewed pro-murder leftist narratives as “moderate,” but “pro-free-speech” narratives as “extreme.”
As I explain, these biases weren’t unique to the UK:
Following Charlie Kirk’s murder, a UK government unit assessed that “Extreme Right-Wing” activists would use Kirk’s assassination to “push pro-free speech narratives” …
@StevenEdginton
spectator.com/article/why-di…
We have all learned from @photomatt and others' experience which is why we now have SSPL and AGPLv3 to stop the egregious leaching. OSI of course doesn't like it. Realistically, these licenses are necessary to maintain commercial investment in open source.
I have nothing to hide from @wpengine in fact I have been quite public with all the things bad about them.
Peter Yared retweeted
The Dutch left-wing government has just decided to revoke the licence of @ongehoordnedtv, the only right-wing TV broadcaster in the public system.
The state is silencing its loudest, and only, critic.
North Korea style.
In een persconferentie licht minister Letschert het besluit om Ongehoord Nederland uit het publieke bestel te zetten toe.
ON!-verslaggever @EljavMiddelkoop: “wat zegt u tegen alle kijkers en leden die nu hun geluid zien verdwijnen van de publieke omroep?”
The trend will be that AI does whack jobs, but let's face it, a lot of unproductive and unnecessary jobs. SEO writing is 100% a BS job.
Does a 3 doctor office really need 10 people at the front desk or are these arbitrary jobs due to nutty regulations and insurance processes? Used to be 1 person at the front desk just 20 years ago.
