𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞

AI, technology & the shifts shaping what comes next.

[Tony Kam @kamikariat]

Excited to launch on @ProductHunt today 😼. Please help upvote 🥺

https://t.co/OzVwaOrwno

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

The real benefit is simple: fewer “Sorry, I didn’t understand that.” moments.

[Shobhit Banga @shobhitbanga]

The world's languages need rigorous speech recognition benchmarks.

Today, @huggingface and @voicearena_ai are adding Hindi and Indian English to the Open ASR Leaderboard, with Monsoon hi and Monsoon en-IN. Each comes with a public split for self-scoring and a private split held out for robust evaluation.

They are part of Monsoon, Voice Arena's data initiative for advancing speech recognition for the world's languages.

Why this matters: benchmarks shape what models get improved. If a language, accent or real-world speaking condition is missing from the benchmark, performance gaps can stay invisible. Most ASR test sets capture what was said, but little about who said it or the conditions in which it was spoken.

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

There’s still so much left for us to discover. This kind of AI won’t do away with science it could dramatically accelerate how fast we make new discoveries.

[Jing @JinghanOng]

Three of our scientists designed a new CAR-T cancer therapy in a matter of months. In preclinical testing, it outperformed a leading therapy used in patients today.

CAR-T reprograms a patient's own immune cells to hunt cancer. Designing one normally takes years and enormous cost.

The difference is what the AI learned from. Not published papers alone, but hundreds of thousands of data points generated from our in-house wetlab runs. Real experimental data, including everything that didn't work.

We're building the superintelligence that takes a scientist from idea to therapy faster than has ever been possible. https://t.co/PJTu0iFTAP

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

@busonix @jun_song This is exactly why I think open-weight AI could get stronger. Nvidia wouldn’t just bring compute it would bring serious institutional weight behind the ecosystem.

[Kiara @ SpaceXAi @kiaraplds]

This is an absolute game changer. Flights, tickets, groceries, anything you need to delegate can be safely handled by your Bot!

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

@ksemenenk0 Багато хто не усвідомлює що за Ші майбутнє. Я сам створюю подібне і бажаю вам успіху у всьому. Це не проста справа тому дивимось тільки в довгострокову перспективу.

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

The foundations of @CrowdStrike approach were sketched on a napkin in a Silicon Valley hotel lounge in 2011.

At a time when betting on cloud-based endpoint security was considered risky, they decided to build around it anyway.

Fifteen years later, they’re training security teams to defend AI systems.

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

On August 31, @CrowdStrike will put security teams through realistic attack scenarios at Fal.Con in Las Vegas.

One of the challenges is entirely focused on AI.

Participants will have to deal with prompt injection, govern shadow agents, and prevent sensitive data from reaching AI models using CrowdStrike Falcon AIDR.

Fal.Con itself will bring together more than 10,000 people from 4,000 organizations across 71 countries. Sponsors include @OpenAI , @AnthropicAI , @googlecloud , @nvidia , and AWS.

More on the Survivor Games here: https://t.co/9m7qZaAILm

We don’t know who exactly will take part in the Survivor Games.

But the interesting part is what happens after them.

https://t.co/0cEAWp8t37

[CrowdStrike @CrowdStrike]

High stakes. No mercy. All fun. The Survivor Games are officially returning to Fal.Con.

On Monday, August 31, defenders across Cloud, IT Ops, SecOps, and SOC teams will enter the arena for an intense showdown of skill, speed, and strategy.

This live-fire test is designed to put your technical expertise and team dynamic to the test against realistic adversary tactics.

If you're heading to Vegas for Fal.Con, join us on Monday! https://t.co/GxoWwro1e9

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

For years, AI helped defenders protect systems.
Now defenders are learning to protect AI itself.

[Robert Scoble @Scobleizer]

We are on our way to Las Vegas via Lake Tahoe and just saw an old guy open up a newspaper stand and take a newspaper out, something I haven't done for 20 years.

Which reminds me: San Francisco, you got to get out of San Francisco once in a while so you can see how fast people change.

It will be a while before we get everybody to change.

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

@Scobleizer There’s something to think about here.

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

Whatever supercar you pick, the fastest thing at Cafe Compute might still be GPT-5.6 Ultrafast on @cerebras.

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

Now this is a nice one.

If you’re heading to Fal.Con in Las Vegas, @cerebras has something interesting for you.

Apply here: https://t.co/yMVwZ027hh

Together with CrowdStrike, they’re hosting Cafe Compute - free food and drinks, AI and cybersecurity people, and a chance to experience Cerebras speed for yourself.

And if you’re lucky, you might even get behind the wheel of a supercar at SPEEDVEGAS. That’s pretty cool.

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

And there’s another interesting part: GPT-5.6 Sol Ultrafast isn’t generally available yet. It’s still in limited preview but at Cafe Compute, you’ll get a chance to experience it for yourself. Not just hear about the speed. Feel it.

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

Now this is a nice one.

If you’re heading to Fal.Con in Las Vegas, @cerebras has something interesting for you.

Apply here: https://t.co/yMVwZ027hh

Together with CrowdStrike, they’re hosting Cafe Compute - free food and drinks, AI and cybersecurity people, and a chance to experience Cerebras speed for yourself.

And if you’re lucky, you might even get behind the wheel of a supercar at SPEEDVEGAS. That’s pretty cool.

[Cerebras @cerebras]

At Cafe Compute Fal.Con, we’re giving one lucky winner the opportunity to experience Cerebras-like speed from behind the wheel of a supercar at SPEEDVEGAS.

The winner take a supercar onto a purpose-built Las Vegas racetrack with a professional instructor riding alongside.

Ferrari, Lamborghini, Porsche, McLaren, Aston Martin, Mercedes-AMG, Nissan GT-R, Lotus. Which car will you drive?

RSVP: https://t.co/9WUFxaJCUL

Come see us at Booth # 1039 at Fal.Con with @CrowdStrike

***
No purchase necessary. One winner will be selected at random at Cafe Compute on Sept 1. Must be in attendance at time of drawing to win. Must be 18+ and have a valid driver's license. Prize is subject to vehicle and reservation availability. Participation may require acceptance of SPEEDVEGAS’s terms and liability waiver. Cerebras is not responsible for personal injury, death, property damage, or other loss arising from participation in or use of the prize.

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

Right now, every task your AI agent runs costs you money. And every file you give it leaves your machine.

Perplexity just changed that. They put the entire agent on your own hardware the planner, the tools, the file index, all of it. Local work costs nothing per token, and nothing is sent anywhere unless you allow it.

Which means it’s finally buyable: an AI that works for you without reporting back to anyone.

If you handle documents, research, or analysis daily, this is where it fits. On Perplexity’s own 53-task test it reached 82.6% on exactly that kind of work. Their numbers, not yet verified by anyone outside.

The cost is $4,699 for the hardware.

And Perplexity is honest about the limit: for demanding coding, the local model doesn’t replace the cloud. Their own test showed it falling behind.

So the shape of it is clear. Your daily work stays home. The hardest thinking still travels.

Where’s your line what would it need to do at home before $4,699 stops feeling like a lot?

[𝐋𝐮𝐜𝐢𝐝 𝐃𝐞𝐥𝐭𝐚 ∞ @LucidDeltaAI]

One day I’ll get one of these and have my own AI team at home. :)

[Robert Scoble @Scobleizer]

“Robotics capabilities could progress a lot faster than the current pace would predict.”