Jorge R. (@JorgeLR_) header
Jorge R. (@JorgeLR_) profile picture
Joined July 2011
Photo by Jorge R. (@JorgeLR_)Photo by Jorge R. (@JorgeLR_)Photo by Jorge R. (@JorgeLR_)Photo by Jorge R. (@JorgeLR_)Photo by Jorge R. (@JorgeLR_)Photo by Jorge R. (@JorgeLR_)Photo by Jorge R. (@JorgeLR_)Photo by Jorge R. (@JorgeLR_)Photo by Jorge R. (@JorgeLR_)Photo by Jorge R. (@JorgeLR_)
$FLOCK ALL YOU NEED!! HODL !!💪
2
1
17
693
Jorge R. retweeted
Bringing private AI to real-world businesses. We’ll be at @GITEX_GLOBAL in Dubai this December as part of the UK Pavilion, organised by @techUK, @biztradegovuk, and Tradefair, showcasing our privacy-first enterprise AI solutions. Who else will be there?
3
1
25
1,406
Jorge R. retweeted
Leave your comments below!
THIS / THAT is already so good I almost feel bad. Almost. Next even larger model is purely made by FLock. Trained completely in-house. It beats you-know-who. yes, I know you guessed it. Question: What should we call it?
1
3
15
1,876
Jorge R. retweeted
One vendor. One outage. Day two, and it's still not fully back. This is what a single point of failure looks like. If your AI lives in the same kind of cloud, what's your plan B? nitter.cf/MSFT365Status/status/2…
We're investigating reports of users experiencing issues accessing Exchange Online mailboxes through multiple connection methods in the Americas. More details can be found in the Microsoft 365 admin center under EX1550543.
3
4
20
1,898
Jorge R. retweeted
The best plan B 🤷‍♀️
One vendor. One outage. Day two, and it's still not fully back. This is what a single point of failure looks like. If your AI lives in the same kind of cloud, what's your plan B? nitter.cf/MSFT365Status/status/2…
2
16
1,575
$FLOCK: IS ANOTHER 5X–20X MACRO RALLY COMING? Previous fractals: ➡️ Dec 2024–Apr 2025: -98% ➡️ Apr–Sep 2025: +1,900% ➡️ Sep 2025–Aug 2026: -96% ➡️ Aug 2026–2027: Potential next expansion? My Accumulation Zone: $0.036–$0.027 (Partial DCA) Targets: $0.077 / $0.19 / $0.69 / $2 Stop Loss: $0.027 (HTF candle close below). 5 BULLISH REASONS FOR @flock_io: 1️⃣ Real decentralized AI infrastructure. 2️⃣ Strong VC backing & enterprise partnerships. 3️⃣ 5.1x AI API usage growth in Q3. 4️⃣ FOMO buyback mechanism creates potential token demand. 5️⃣ ~$20M market cap offers significant growth potential. My macro expectation is a potential 5X–20X if historical fractals repeat and execution delivers. I am bullish, but not chasing green candles. Accumulation with disciplined risk management is my approach. High potential, but execution is everything. NFA. DYOR. #FLOCK
13
12
137
12,584
Sarawak, Malaysia’s largest state, is moving first on sovereign AI in Southeast Asia, and we’ve been part of the build with Sarawak AI Centre. This week, SoftBank, Grab and PETROS are putting $1B into AI infrastructure there, part of $1.85B for Malaysia. But sovereignty isn’t only about where compute sits. It’s also about how models are trained. We're building that layer with federated learning and distributed AI, so government bodies and organisations can train their own specialised models and keep full control of their data.
1
5
273
Jorge R. retweeted
Sarawak, Malaysia’s largest state, is moving first on sovereign AI in Southeast Asia, and we’ve been part of the build with Sarawak AI Centre. This week, SoftBank, Grab and PETROS are putting $1B into AI infrastructure there, part of $1.85B for Malaysia. But sovereignty isn’t only about where compute sits. It’s also about how models are trained. We're building that layer with federated learning and distributed AI, so government bodies and organisations can train their own specialised models and keep full control of their data.
I welcome the announcement by the SoftBank-led consortium, including Grab and PETROS, to invest in AI Infrastructure Platform in Sarawak and Peninsular Malaysia. This is a significant moment in the country’s digital transformation and reflects global confidence in Malaysia’s potential to develop and use AI. It supports our ambition to become an AI Nation by 2030, as set out in the National AI Action Plan 2026–2030. The government will work with the relevant state authorities to facilitate this highly anticipated project. The willingness of global investors like SoftBank and Masayoshi Son to invest in Malaysia reflects strong confidence in our national capabilities, investment climate and political stability. Integrating SoftBank’s capital and expertise with Grab’s regional AI capabilities and proven track record of deploying operational AI across its ecosystem will significantly enhance Malaysia’s AI infrastructure and global competitiveness. As Malaysia develops its AI infrastructure, we expect large-scale digital investments to support our sustainability goals. This project must meet applicable federal and state requirements, including those on renewable energy use. I expect this investment to create skilled jobs for Malaysians and help more of our rakyat build expertise in AI. It should also create opportunities for our local businesses to supply goods and services as Malaysia builds its AI capabilities. #MalaysiaMADANI #YakinMADANI
1
5
1
29
2,520
Jorge R. retweeted
Our Chief AI Scientist @zehuac was in Rome to present six FLock papers in the @CIKM 2026 Full Research Track. Together, they map out what we need to make private AI training affordable, safe, and reliable at scale. 1. Cheaper training Most training updates the whole model at once, even though the parts affect each other. Gradient Interaction Graph Optimization updates them in a smarter order and reaches the same quality in about a third fewer steps. 2. Safer model merging Combining two models can work well or ruin both, and normally you only find out after running expensive tests. Curvature Graph Mergeability predicts the outcome in advance from a quick scan of each model. 3. Robustness to bad examples Showing misleading examples in a prompt can throw off the whole answer. In-Context Learning as Belief Propagation lets the model recognise the odd one out and trust it less, improving accuracy by about 7 points. 4. Keep structured outputs consistent When AI fills in a form, it often decides each field separately, so a junior job title can end up next to a senior salary. Bethe Consensus Decoding checks the fields against each other and returns a consistent answer, with no retraining. 5. Quantify uncertainty A model should be able to say “it’s one of these three” when it isn’t sure. Cavity-Calibrated Prediction Sets give guaranteed-reliable shortlists that are up to 58% tighter, so they stay useful. 6. Anonymise training text Real-world text is full of names and personal information that regulations keep off-limits. Graph-Driven Context-Preserving Anonymization removes 99.8% of it while keeping the text just as useful for training. Private AI becomes more critical as data becomes every organisation’s edge, and FLock is building at that frontier.
1
6
1
35
2,172
Every government I've spoken to wants AI in public services. The hesitation is never about the technology. It's about the data: where it goes, who can see it, and what happens if it leaks. That's why we start from the other end. Keep the data where it is, and bring the AI to it. I shared how we take this from pilot to scale in techUK's Digital Transformation in the Public Sector series. Thanks @techUK for having me.
The UK is home to some of the world’s best AI research. Now it needs more AI companies that turn ideas into revenue and stay here. Three ways to get there: 1. Keep compute spend at home: Pooled, distributed compute keeps AI R&D funding in the UK instead of paying GPU rent to overseas hyperscalers. 2. Use privacy to unlock scale: Federated learning lets hospitals, councils and banks train their own specialised models on proprietary data without it leaving the building. 3. Back the step from pilot to production: Favourable policies and the public sector as an early customer would help far more UK teams land a first contract. Our Head of EMEA @TiffanyWang98 shared these on the “Buy, Back, or Get Out of the Way” panel at @techUK’s Tech & Innovation Summit 2026. Thanks for having us.
5
10
417
Jorge R. retweeted
When you look at this that 1.7 You will know how amazing FLock is.
1
4
2
21
1,771
Jorge R. retweeted
FLock amazing 🚀🚀🚀 $$$ will be soon
When you look at this that 1.7 You will know how amazing FLock is.
1
1
41
Jorge R. retweeted
1.5 multi modal Any more I can say? @zehuac Or surprise everyone.
2
3
1
15
1,086
Jorge R. retweeted
Every great migration starts with a leap of faith. @flock_io is moving $FLOCK and FOMO Model Tokens onchain liquidity to @Uniswap on @base, and @RobinhoodCrypto’s Robinhood Chain. Fly with the unicorns. Fly higher 🦄
5
8
2
43
2,865
Jorge R. retweeted
Flying with the unicorns 🦄
1
3
2
27
2,139
Jorge R. retweeted
Most enterprise AI still starts with a tool. Here's the order we think it should run in. Leadership sets the ambition. Then engineers sit with the people who'll use the system, map how a decision gets made today, and only then decide what to build. Everything is built inside the organisation's own environment, so the data, models and tools stay theirs. Success is measured on business metrics from day one. The AI advises. The professional makes the call. It's how FLock works with hospitals, law firms and governments.
3
2
1
25
2,252
Jorge R. retweeted
FLock is flock awesome🔥🔥🔥
Q3 at FLock, in numbers: - 11 research papers accepted at top AI and data-mining conferences - 2.5M+ decisions processed by THIS / THAT Model in 3 weeks - FOMO goes cross-chain - New milestones on FLock API Platform, AI Arena and institutional adoption Take a closer look ↓
1
5
93
Jorge R. retweeted
Absolutely flocked in over the last 3 months: → @RedHat ecosystem partner → Collaboration with UNDP Dominican Republic → THIS / THAT Model: ultra-fast decision model by FLock with 1.88B params, 2.5M+ decisions in its first 3 weeks → 11 research papers accepted at top AI/ML conferences CIKM, ECML PKDD and IEEE ICDM → FLock's FL case study in healthcare got featured on @wef's MINDS Programme → 5x tokens processed on the FLock API Platform (from 5.4B in July to 27.7B in September) → $FLOCK goes cross-chain via @chainlink CCIP, perps live on @okx Q3 @flock_io ⤵️
3
5
3
35
3,421
Q3 at FLock, in numbers: - 11 research papers accepted at top AI and data-mining conferences - 2.5M+ decisions processed by THIS / THAT Model in 3 weeks - FOMO goes cross-chain - New milestones on FLock API Platform, AI Arena and institutional adoption Take a closer look ↓
1
153
Jorge R. retweeted
Q3 at FLock, in numbers: - 11 research papers accepted at top AI and data-mining conferences - 2.5M+ decisions processed by THIS / THAT Model in 3 weeks - FOMO goes cross-chain - New milestones on FLock API Platform, AI Arena and institutional adoption Take a closer look ↓
5
7
3
34
3,532