Emma

[M. Brandon Lee | THIS IS TECH TODAY @thisistechtoday]

Heads up, MicroCenter has the base M6 Apple Mac Mini on sale for $699.99.

Best Buy price matches it.

If you have the M4 Mac Mini, you can get $500 for it.

I just did it!

Easy $200 + tax upgrade. Not bad!

[Anthony Kroeger @kr0der]

bro Cloudflare just launched like 100000 products in the past week

i feel like you’re gonna be able to ask your AI agent to use the cf CLI and set up literally everything for your product on Cloudflare

super bullish, i love Cloudflare man

[Emma @Emma12379127]

@HamzaLamoudni Apps

[David Hendrickson @TeksEdge]

👀 You guys might not have seen this Ryzen AI Max+ Pro 495 box yet.

And this one is a little different.

Lucebox Zero 495
💰 $8,798 192GB version, first-200 pricing
🧠 192GB LPDDR5X-8533 @ 273 GB/s
🎮 32GB Radeon AI PRO R9700 @ 640 GB/s
💾 2TB NVMe
🌐 5GbE
🚀 optional 100GbE

So unlike most of the new 192GB Gorgon Halo boxes I've posted…

this one also has 32GB of fast dedicated VRAM.

That's 🧠 192GB Halo memory

⚡ 32GB R9700 VRAM = 224GB across two memory pools

And Lucebox built its inference engine specifically to use both.

Smaller/hot workloads can live on the faster R9700, while much larger models can use the huge Halo memory pool.

They say a single model can also be split across both.

⚠️ This is not one giant 224GB VRAM pool. They're separate memory pools managed by the software.

🔗 Link in ALT.

[Emma @Emma12379127]

@RCIllingworth @emailoftoday Stack

[Marco Franzon @mfranz_on]

NVIDIA Spark 64 GB $4999

AMD AI Halo 128 GB $3999

Could you explain me what is happening please?

[0xSero @0xSero]

Best deal without a doubt right now grab it before it’s too late.

With this you’ll be able to run GLM-Flash/DS4.1-Flash really fast

❤️

[Emma @Emma12379127]

@CopyWithOlli Copy

[Tony Simons @tonysimons_]

Four AI agents. Four VERY different answers to one question:

Who gets the keys?

I broke down OpenAI dots, Grok Bot, Meta Muse, and Hermes Agent on control, permissions, pricing, privacy, and what they’re actually built to do.

No “smartest AI” nonsense.

👇

https://t.co/BvPfaGXKcp

[Nandkishor @devops_nk]

If you want to become a DevOps engineer, learn Linux first.

You don’t need 1000 commands.

You need to understand the right commands and how to troubleshoot with them.

This Linux cheat sheet covers:

• File management
• Permissions
• Processes
• CPU, RAM & Disk monitoring
• Systemd & services
• Networking
• SSH troubleshooting
• Logs & environment variables
• Package management
• Archives & compression
• Essential DevOps commands
• Troubleshooting flow

🧠 The Linux troubleshooting mindset:
Check → Identify → Investigate → Fix → Verify → Monitor

Save this for your next Linux revision.

Linux skills compound. Practice every day.

[Saurav Chaudhary @sauravstwt]

🚨 9 things I wish someone told me before I became a DevOps engineer.

1. Linux matters more than your 14th DevOps tool.
2. Networking will humble you eventually.
3. Kubernetes doesn’t eliminate distributed-systems problems.
4. Monitoring isn’t observability.
5. Automation can automate disasters too.
6. The database is part of your infrastructure problem.
7. A rollback strategy matters more than a beautiful deployment pipeline.
8. The best incident responders aren’t command encyclopedias. They reduce the search space faster.
9. Nobody cares how beautiful your YAML is when checkout is down.

Learn the tools. But learn why systems fail even harder.

Think Beyond YAML.
🔁 Repost If you find this list useful !

[Stefan Georgi @StefanGeorgi]

DTC Secret #13 - Leverage Microvariants

Okay so whenever you write a new ad…

You don’t just write one single hook.

Instead, you write your primary hook…

And then you create 5 to 7 additional hooks and opening lines for that exact same ad.

That way, your media buyers and editors have multiple variations to test right out of the gate.

This is standard operating procedure…

And yet, this is also where so many marketers (and especially AI) get stuck.

Because when it comes time to write those 5 to 7 alternative hooks…

People usually make one of two big mistakes (and they are very much related):

**Mistake #1: Pulling Ad Copy to the Top and Calling It a Hook

This mistake gets made by both humans and AI…

But I’ll focus on the AI part of the process since most folks use LLMs for their copy these days.

Let’s say you’re writing an ad for a skincare product…

The concept is about how heavy metals can damage your skin…

And the product has ingredients that help your skin cells detox those heavy metals (a process called chelation).

You write the ad..

Then you ask the LLM for alternative hooks…

And it just crawls deeper into the body copy of the ad and then pulls out that copy and pretends it’s a hook.

So instead of a fresh opening, it spits out boring, mechanism-heavy stuff like:

“Here is how chelation binds to heavy metal ions to protect your collagen.”

Or it just takes some moment that happens later in the ad and pulls it to the front, like this:

“A friend told me about this eye serum at brunch, and after two weeks I was shocked by the results.”

The first hook sucks.

The second hook is not bad… the FIRST time you see it.

But when AI just keeps giving you that same hook again and again, it gets frustrating fast.

**Mistake #2: The “Net New” Mistake

Both marketers and LLMs typically assume that every alternative hook needs to be a completely different, unrelated premise.

So say your winning hook is something like the following:

“I washed my face with filtered water for 30 days, and the results were crazy.”

Okay, that’s a decent hook…

But what could those alternative hooks you create for the ad look like?

Typically, if you ask AI, or a junior creative strategist…

It’ll be stuff like I mentioned in Mistake #1:

“Here’s what heavy metals do to your skin.”

“My dermatologist just told me about something called chelation, and I need to tell someone about it.”

Stuff like that.

I’m not saying those hooks are horrible…

It’s just that they ignore some of the lowest-hanging fruit when it comes to your ads…

Which is to test microvariations of the initial hook.

And that looks like this:

Original Hook:

“I washed my face with filtered water for 30 days, and the results were crazy.”

Microvariant 1: “I washed my face with refrigerator water for 30 days, and the results were crazy.”

Microvariant 2: “I washed my face with bottled water for 30 days, and the results were crazy.”

Microvariant 3: “I washed my face with purified water for 30 days, and the results were crazy.”

Microvariant 4: “I washed my face with ice-cold mineral water for 30 days, and the results were crazy.”

Microvariant 5: “I washed my face with bottled mineral water for 30 days, and the results were crazy.”

Microvariant 6: “I washed my face with volcanic mineral water for 30 days, and the results were crazy.”

See what we did there?

We took one decently strong hook and turned it into 6+ hooks we could test.

And, honestly, I could keep going…

“Glacier Water,” “pH Water,” “Alkaline Water”

The list goes on and on.

And when you do this…

You will be shocked by just how big of a difference changing a few words in your hook can make.

But also, it makes sense…

Because while “filtered water” might not grab a prospect’s attention and stop the scroll…

“Volcanic mineral water” very well may.

-SPG

P.S. A quick pro tip if you’re using AI to help brainstorm:

When asking an LLM for alternative hooks, don’t just say: “Give me X more hooks for this ad.”

If you do that, it’ll default to Mistake #1 and give you robotic mechanism lines.

Instead, tell it:

“Give me X micro-variations on this opening line by swapping out the core scenario or noun with more unusual, curiosity-inducing alternatives.”

You’ll tend to get some interesting results.

For example, I just took that initial hook and asked StefanBrain to give me 5 more variations, and here’s what I got:

“I washed my face with sparkling mineral water for 30 days, and the results were crazy.”

“I put grass-fed beef tallow on my face every night for 30 days, and the results were crazy.”

“I washed my face with medical-grade saline for 30 days, and the results were crazy.”

“I washed my face with fermented rice water for 30 days, and the results were crazy.”

“I dunked my face in a bowl of ice water every morning for 30 days, and the results were crazy.”

Plus it gave me this bonus one, which is kind of interesting:

“I washed my face with raw unpasteurized honey for 30 days, and the results were crazy.”

[Jordan @jordanpaid]

I genuinely believe that if you have a mindset of wanting to serve others, you’ll always do well in business

At the end of the day, that’s what business really comes down to.

You’re solving people’s problems, you’re helping people, you’re creating products and services that make someone’s life better in some way, shape or form

Don’t go into business thinking about how much you can take, go into it thinking about how much value you can give, then I believe you’ll always be rewarded

[Jordan @jordanpaid]

One thing that really motivates me is that i don't want to look back in 10 years and think I played it too safe

that's probably the biggest thing driving me right now

i'd rather try something and fail than spend months wondering what would've happened if I actually WENT for it.

You get one shot at this life

Might as well see how far you can take it

[Nandkishor @devops_nk]

Raven is basically bringing the AI agent team model to DevOps.

Imagine one team handling research, coding, design and on-call while continuously improving its own playbooks.

DevOps is slowly moving from managing infrastructure to orchestrating AI that manages the infrastructure.

[Jun Song @jun_song]

I called this a long time ago.

People will choose local AI with no limits, even if it is slightly less smart. The truth is, it is already good enough for most real work.

Consumer hardware prices are going to go crazy.

[Nandkishor @devops_nk]

As a DevOps Engineer who’s worked with AWS/Azure, Deployment and Auto scaling InstaCloud caught my attention.

It’s built for AI agents they can deploy, create replica environments, test, and autoscale while InstaCloud handles the infrastructure.

They’ve raised $8M. Curious to see how this plays out against traditional cloud providers like AWS and AZURE.

[Cristian Córdova 🐧 @barckcode]

Si son SysAdmins/SREs/DevOps/Platform Engineers/[Deje aquí su role fancy de infra].

Aprendan sobre las profundidades de la IA y cómo correr y escalar modelos. GPUs, CUDA, inferencia, caché, etc

Su economía de los próximos años se lo agradecerá enormemente…

[Marino Wijay 🇨🇦 @virtualized6ix]

This is your friendly reminder that working with infrastructure is still DAMN HOT🥵🔥

There’s still plenty of time to learn this stuff. A combo of these areas will make you DANGEROUS😎:

- NETWORKING
- Linux
- Kubernetes, containers
- Storage and data management
- Virtualization/Cloud
- Monitoring, observability, telemetry, alerting
- Automation, IaC, CI/CD, git
- vulnerability management, segmentation, policy deployment/enforcement
- BCDR

[Nandkishor @devops_nk]

We’re so back.

- DevOps Engineer
- Platform Engineer
- Infrastructure Engineer
- SRE Engineers

will be needed.