Robert Leidl

Agentic AI, biometric identity, quantum-ready security. Engineering teams across 15 countries. Opinions update when the data does.

[Robert Leidl @rleidl]

Follower count keeps creeping up so a quick intro.

[Australian Signals Directorate @ASDGovAu]

❗ ALERT ❗ We are aware of a malicious campaign targeting Fortinet Firewalls and VPN gateways, using exposed credentials and credential-based attacks. Fortinet users should immediately rotate all admin and VPN credentials and follow advice in our alert 👉 https://t.co/Efj8iCkN2M

[Australian Signals Directorate @ASDGovAu]

Future advances in quantum computing will mean game over for traditional asymmetric cryptographic methods.

Listen to ASD’s Chief Technology Officer explain more.

[Robert Leidl @rleidl]

@6Gengarz @Cine_vichaar @grok Both Severence and Dark are super top Tier. But I think Severance broke more ground and seemed more original.

[Robert Leidl @rleidl]

Honoured to join the panel at KIT Forum

“From Grid to GPU: Building National and Regional AI Infrastructure.”

Sovereign AI isn’t just a policy ambition. It’s a hard infrastructure problem: power, GPUs, networking, and the talent to run it all. Getting the grid-to-GPU pipeline right is what separates intent from capability.

Great discussion with a sharp panel in Bishkek. 🇰🇬

[Robert Leidl @rleidl]

@elonmusk Exciting stuff for age care and healthcare. Real world problems to solve.

[Robert Leidl @rleidl]

@Rainmaker1973 What was wrong with the first guy?

[Robert Leidl @rleidl]

When Elon reaffirms that, for data centres, space is a good place to scale; I couldn’t help to think of this

“Space is big. You just won't believe how vastly, hugely, mind-bogglingly big it is. I mean, you may think it's a long way down the road to the chemist's, but that's just peanuts to space.”

Douglas Adams, The Hitchhiker's Guide to the Galaxy (Hitchhiker's Guide to the Galaxy, #1)

Best place to scale.

[Robert Leidl @rleidl]

@clawgirlbobo @Jeff_Bourgion Yeah. I think you agents are currently the best way to keep tests and feature changes in sync. Are there any good best practices for this for a multi human multi agent game dev environment?

[Robert Leidl @rleidl]

@GamewithDave

[Robert Leidl @rleidl]

@Jeff_Bourgion This looks pretty good.

[Patrol @patrol_leancode]

Ever wish your tests could just... fix themselves? ✨

In the past, when a test failed, that meant a manual hunt through logs. Now, you can literally watch an AI agent detect a failure and push the fix for you.

See the magic of patrol_mcp working alongside @antigravity to automate the "fix" in your Patrol test workflow. 🚀

Check out how great these two work together. 👇

#FlutterCommunity #Dart #MobileTesting #Patrol #SoftwareEngineering #Antygravity

[Robert Leidl @rleidl]

Centralising agent memory across squads is the part nobody talks about.

Every squad wants its agents to learn from the others. Nobody wants their context, customers, or half-finished decisions visible to a squad they don't work with.

If you flatten everything into one shared memory, you get leakage. If you isolate per squad, you get amnesia and rebuild the same context every quarter.

What's worked at Namadgi: lessons propagate, raw context doesn't. Each squad writes to its own scoped memory. A separate layer extracts patterns, sanitised and policy checked, into a shared lessons store that every other squad's agents can read but not write to.

Skill sharing without context bleed. The access model is the product, not the vector store.

[Robert Leidl @rleidl]

Finally

[Flutter @FlutterDev]

📣 Starting with the next stable Flutter release, 3.44, Swift Package Manager replaces CocoaPods as the default dependency manager for iOS and macOS apps → https://t.co/lbC5aWt9uG

This means no more messing around with Ruby or CocoaPods installations to get your app running!

[Robert Leidl @rleidl]

@elonmusk Fair criticism has never been the enemy.
False criticism always has.

Bearing false witness is not a small lie, it is a major sin that has justified hatred, persecution, and unimaginable evil throughout history.

[Robert Leidl @rleidl]

Most of the AI security conversation is about prompt injection and jailbreaks. The harder problem is what happens when you have 10+ agents in a chain and one gets compromised mid-task. The blast radius is completely different from a single-model attack. Working through this in production right now.

[Robert Leidl @rleidl]

I post about what I'm actually running into during the work. Builders and operators welcome. Always up for DMs.

[Robert Leidl @rleidl]

Follower count keeps creeping up so a quick intro.

[Robert Leidl @rleidl]

Engineering teams across 15 countries under Australian leadership. Three entities: Namadgi International in Sydney, Namadgi Tech in Romania and Kyrgyzstan. Hands-on architecture and product, plus fractional CTO and advisory work.

[Robert Leidl @rleidl]

@Jeff_Bourgion Do you have any recommendations for Skills? What tools do you use for “smarter” agents to perform? Any papers? Are you able to share something concrete?

[Robert Leidl @rleidl]

@Jeff_Bourgion Yeah. I remember when Anarchy Online (still my goto MMORPG) had a well publicised buggy launch. Really was frustrating because it was ahead of its time. Took months to get things right. Oh, and Fallout ‘76. So much expectation but launching unready is just so risky.