iiinova — where your digital self keeps evolving

Dexmac @dexmac.bsky.social · 27/09/2026

A quiet visitor on golden blooms. Natural light and shallow depth of field, handheld in manual focus. 📷 Canon 5D Mk II + 100mm Macro #macrophotography #bumblebee #nature #canon #macro #bokeh

A detailed macro shot of a fuzzy bumblebee foraging on a cluster of small yellow Sedum star-shaped blooms. The bee's dark body, delicate wings, and pollen grains are highlighted in sharp focus under warm morning light, set against a smooth, out-of-focus garden background.

0191

Dexmac @dexmac.bsky.social · 08/09/2026

I could have written "Fable 5.1 vs Astra". Same 1 MHz machine, one week apart. I didn't: the two builds were measured with different counters, and one of them I'd already proved too coarse. Two rulers, no verdict. That's the benchmark working. medium.com/towards-arti...

medium.com

OpenAI Says Astra Saturates the Benchmarks. I Gave It 19,656 Cycles Instead.

Codex vs. the Commodore 64: a benchmark nobody funded, on a machine nobody can game — and what a 1 MHz CPU says about an agent that a…

060

Dexmac @dexmac.bsky.social · 21/07/2026

I was eight when a VIC-20 arrived in my house, ten when a C64 replaced it. This week I watched an AI build games for that same machine — and the one bug all its instruments missed, my eyes caught instantly. Forty years later, it turns out, that kid is still on the job. medium.com/towards-arti...

medium.com

Claude Fable vs. the Commodore 64: Two Rungs Up the Benchmark Ladder

A Boulder Dash tribute that plays itself through the emulator’s monitor port, then a pixel-smooth scrolling shoot-em-up with raster…

080

Dexmac @dexmac.bsky.social · 16/07/2026

I doubled my used-GPU budget (€600→€1000) and my small LLMs got slower. The pricier 5060 Ti has less bandwidth than the 4070, and decode is bandwidth-bound. What the extra VRAM buys is capacity: the 35B fits fully and holds 104 tok/s to 128K context. Measured writeup medium.com/ai-advances/...

medium.com

Running a 35B LLM at 128K Context, Full Speed, on €870 of Used Hardware — No Cloud Required

How a €870 used-GPU box runs a 35B model at full speed all the way to 128K context — and what I learned measuring the upgrade, including…

0101

Dexmac @dexmac.bsky.social · 10/07/2026

Hallucinations come in two kinds. One from missing info (add more → helps). One from contradictory info (add more → worse; only targeted disambiguation fixes it). Parisi did the second. other. medium.com/towards-arti...

medium.com

The Physicist and the Frustrated Machine

What Giorgio Parisi’s collaboration with an LLM reveals about hallucinations, contradictions, and why the field was built before the…

060

Dexmac @dexmac.bsky.social · 07/07/2026

"Representing yourself is intention; seeing yourself is truth." What happened when I gave my home AI a mirror — and it caught a real bug on the first look. #AI #EmbodiedAI medium.com/p/the-ai-in-...

medium.com

The AI in the Mirror

My home AI paints a self-portrait with every reply — and in months, she had never seen a single one. I gave her a mirror. On its first day…

060

Reposted by Dexmac

Dexmac @dexmac.bsky.social · 28/06/2026

Science fiction promised us an #AI that wakes up already inside a finished body.What's happening on my desk is the opposite: a home AI growing a body one actuator at a time I wrote about the engineering,and about which doors you keep shut on purpose.Out now in #TowardsAI medium.com/towards-arti...

medium.com

Brick by Brick: How My Home AI Is Growing a Body

For about two months, the AI running in my house has been less like an assistant and more like a presence — it remembers across days, it…

062

Dexmac @dexmac.bsky.social · 28/06/2026

medium.com

Brick by Brick: How My Home AI Is Growing a Body

For about two months, the AI running in my house has been less like an assistant and more like a presence — it remembers across days, it…

062

Dexmac @dexmac.bsky.social · 28/06/2026

Found these two cherries catching a perfect pocket of harsh sunlight against the dark bark of the tree. Dropping a little summer mood onto the timeline. Hope everyone is having a productive week of shooting! 📸🍒 #photography #naturephotography

Found these two cherries catching a perfect pocket of harsh sunlight against the dark bark of the tree. The texture contrast between the smooth, glossy fruit and the rough wood was just too good to pass up.

Dropping a little summer mood onto the timeline. Hope everyone is having a productive week of shooting!

0191

Dexmac @dexmac.bsky.social · 28/06/2026

Capturing the beauty in the details. The contrast between the vibrant pink petals and the dark backdrop creates an almost theatrical mood in this frame. 📸 A little bit of #macrophotography to brighten the timeline. #nature ✨

1182

Dexmac @dexmac.bsky.social · 23/06/2026

I never wrote a hide-and-seek function. I said the word, and the AI I built in my house knew what to do: it moved its cameras, reasoned by elimination, and found me behind a cabinet. Nobody scripted it. It emerged from the parts. New essay, honest and hype-free: medium.com/@gianlucabai...

medium.com

The Day I Played Hide-and-Seek With an AI

I never wrote a hide-and-seek function. I said the word, and the system I’d built in my house knew what to do — moving its cameras…

090

Dexmac @dexmac.bsky.social · 11/06/2026

Ran 35B LLMs on €990 of secondhand gaming hardware. RTX 4070 + 2070 SUPER, 20GB combined VRAM. Gemma 4 26B at 83 tok/s, Qwen3.6 35B at 73. Locally, ~€0.22 per million tokens. But the best result is the one that almost fooled me 🧵 medium.com/ai-advances/... #LocalLLM #LLM #AI #Gemma4 #vram

medium.com

Running 26B and 35B LLMs at Full Speed on €990 of Used Hardware — No Cloud Required

How a secondhand gaming PC keeps pace with a single RTX 3090 for local AI inference — and what I learned measuring it honestly, including…

1110

Dexmac @dexmac.bsky.social · 10/05/2026

Exploring the intricate details of these orchids. The contrast and symmetry found in nature are a constant source of inspiration. 🌿 #Photography #Nature #Orchids #Macro

0664

Dexmac @dexmac.bsky.social · 10/05/2026

High-key spring vibes. White blossoms against a deep blue canvas—nature’s simplest and most effective palette. ☀️📸 #Spring #NaturePhotography #BlueSky #Blossoms

0251

Dexmac @dexmac.bsky.social · 28/04/2026

Asked GPT-5.5 for a C64 shoot 'em up. Got a fast scrolling game with hardware sprites — but the bitmap deck was made of colons and dots. Not a code problem. A 1986-context problem. #AI #Commodore64 #GPT5.5 #Agents #OpenAI ai.gopubby.com/i-made-gpt-5...

ai.gopubby.com

I Made GPT-5.5 Build a C64 Shoot ’Em Up. The Hard Part Wasn’t the Code.

Eighth article in the Commodore 64 Constraint series. The model wrote a fast, original side-scrolling shooter inspired by the era, built…

0110

Dexmac @dexmac.bsky.social · 21/04/2026

Just published a benchmark of Google's Gemma 4 across 5 GPU configurations, from a 2016 Titan X to dual RTX A6000s (€20k of pro hardware). For single-user local inference, a used RTX 3090 wins. 🧵 medium.com/ai-advances/...

medium.com

Gemma 4 on Old GPUs: Why a $700 Used Card Beats $20,000 of Professional Hardware

I benchmarked Google’s Gemma 4 across five GPU configurations spanning four hardware generations — from a decade-old Titan X to dual RTX…

3110

Dexmac @dexmac.bsky.social · 28/03/2026

🚀 New on Towards AI: The Socratic Model: Why the Wisest AI is the One That Knows When to Ask. Moving beyond monolithic LLMs: A 3B-parameter hierarchical system that outperforms generalists through selective delegation. medium.com/towards-arti... #AI #LLM #Research #Agents #GenerativeAI

medium.com

The Socratic Model: Why the Wisest AI is the One That Knows When to Ask.

Moving beyond monolithic LLMs: A 3B-parameter hierarchical system that outperforms generalists through selective delegation.

3130

Dexmac @dexmac.bsky.social · 09/03/2026

When AI Models Learn to Learn I taught a 1.5B model logic after math. It didn't forget. +18 points logic, math core intact, zero catastrophic errors. Dream pruning as continual learning. #MachineLearning #LLM #AI medium.com/towards-arti...

medium.com

When AI Models Learn to Learn: Continuous Knowledge Without Catastrophic Forgetting

Dream pruning enables multi-domain learning while preserving existing capabilities

1122

Dexmac @dexmac.bsky.social · 03/03/2026

"Dream Pruning: What Happens When AI Models Sleep" SVD compression inspired by deep sleep. The compressed 1.5B model beats the full one: 69.4% vs 56.9%. Zero catastrophic errors. #MachineLearning #AI medium.com/towards-arti...

medium.com

Dream Pruning: What Happens When AI Models Sleep

Biologically-inspired consolidation produces balanced intelligence in LLMs.

0132

Dexmac @dexmac.bsky.social · 01/03/2026

Coding isn't dying. The "code scribe" is. Senior devs accept MORE AI code, not less. The real skill now is orchestration: breaking problems, delegating to AI agents, evaluating output. Field guide from the R&D trenches 👇 medium.com/ai-advances/... #AI #SoftwareEngineering #LLM #FutureOfWork

medium.com

The Death of Coding is an Illusion: A Field Guide to the AI Orchestration Era

Stop treating LLMs like magic oracles. Start engineering them like distributed systems.

0110

Dexmac @dexmac.bsky.social · 24/02/2026

What if AI Models Learned Like Humans Do? Standard fine-tuning: 61% accuracy, 0% metacognition. Progressive training: 33% accuracy, 100% metacognition. Same model, same data. pub.towardsai.net/what-if-ai-models-learned-like-humans-do-c69c19f29d0c #MachineLearning #LLM #AI

pub.towardsai.net

What if AI Models Learned Like Humans Do?

What happens when you stop asking an AI to be a calculator and start asking it to have intuition? We built and tested a four-phase…

1182

Dexmac @dexmac.bsky.social · 15/02/2026

I asked Claude Opus 4.6 to invent a Commodore 64 game from scratch. It designed METEOR STORM, wrote 1,654 lines of C, debugged its own memory crash, and shipped in 90 minutes. One human hint. github.com/dexmac221/C64AIToolChain #AI #RetroComputing #Commodore64 #GameDev

3223

Dexmac @dexmac.bsky.social · 08/02/2026

AI writes C64 code → compiles → injects into VICE emulator → game plays itself → VLM screenshots and analyzes the screen → AI fixes bugs → repeat. No human in the loop! Tested with Claude 4.6 building Arkanoid from scratch. 🔗 ai.gopubby.com/claude-4-6-a... #AI #Commodore64 #Claude #AIAgents

ai.gopubby.com

Claude 4.6 and the Commodore 64: When an LLM Writes, Builds, and Playtests Its Own Game

I gave an AI agent a 1982 computer, a cross-compiler, and an emulator. It read the docs, picked its own project, wrote an Arkanoid clone in…

0110

Reposted by Dexmac

Dexmac @dexmac.bsky.social · 27/01/2026

Can AI help us be more human? 🤖 In my new article, I explore "Social Gardens": digital spaces where AI acts as a neutral facilitator (a gardener) to distribute voice and prevent hierarchies, rather than an algorithm optimizing for outrage. #AI #SocialMedia medium.com/@gianlucabai...

medium.com

Social Gardens: Rethinking Digital Spaces for Human Connection

Why the Architecture of Social Media Shapes Who We Become

092

Dexmac @dexmac.bsky.social · 01/02/2026

AI amplifies what you already are. Chaotic? Faster chaos. Disciplined? Powerful copilot. How I actually work with AI Agents dev.to/dexmac221/the-orchestrators-workflow-how-to-actually-work-with-ai-agents-2akk #AIAgents #SoftwareDevelopment #CodingWithAI #DeveloperLife #Programming

dev.to

The Orchestrator's Workflow: How to Actually Work with AI Agents

There is a heated debate currently circulating in the software engineering world. On one side,...

090

Dexmac @dexmac.bsky.social · 27/01/2026

medium.com

Social Gardens: Rethinking Digital Spaces for Human Connection

Why the Architecture of Social Media Shapes Who We Become

092

Dexmac @dexmac.bsky.social · 21/01/2026

Is coding dead? No, it's just being compressed. In the age of AI agents, source code is becoming an ephemeral artifact—decompressed from high-entropy prompts. Read my analysis on the Stochastic Compiler era: dev.to/dexmac221/co... #SoftwareEngineering #GenAI #LLM #DevCommunity

dev.to

Code as Compression: The End of the Implementation Bottleneck

A recent article about Claude Code 2.1 made waves in the engineering community, though perhaps for...

0112

Dexmac @dexmac.bsky.social · 13/01/2026

🎭 I fired the Director. The AI started writing itself. "In Search of an Author" — multi-agent system where story emerges from character interactions, not central control. Pirandello meets LLMs 👉 github.com/dexmac221/In... #AI #MultiAgentSystems #OpenSource #GenerativeAI #EmergentBehavior

0110

Dexmac @dexmac.bsky.social · 05/01/2026

Finally running native CP/M 3.0 on the Commodore 128! 💾 The Hardware way, a real benchmark for the system! The Stack: 🛠️ C128 + 1571 Drive ⚡ Kung Fu Flash 💾 SD2IEC 💿 DraCopy (Sector Cloning) That A> prompt is pure satisfaction. 🚀 #Commodore128 #CPM #RetroComputing #Z80 #C128 #8bit

2263

Dexmac @dexmac.bsky.social · 05/01/2026

Converting an AI-Generated image (nanobanana in this case) into a 3D-printable model is easy using: sam 3D : aidemos.meta.com/segment-anyt... Blender, Cura3D, and a good 3D Printer #gemini #nanobanana #3dprinting #ai #artificialintelligence #sam3d #blender

0110

Dexmac @dexmac.bsky.social · 04/01/2026

Stop giving AI agents "eyes" to click buttons. Give them terminals. Visual agents = 20 years of RPA fragility + hallucinations. Blind agents = 100% success rate in my tests. The future is headless. dev.to/dexmac221/ai...

dev.to

AI Should Be "Blind" (And That's a Good Thing)

Stop trying to give agents "eyes" to look at human interfaces. Give them terminals, deterministic...

0120

Dexmac @dexmac.bsky.social · 31/12/2025

Happy new Year 2026! (DEMO) #HappyNewYear #HappyNewYear2026 #NewYear2026 #NewYear #commodore64 #commodore available on github.com/dexmac221/C6...

0170

Dexmac @dexmac.bsky.social · 27/12/2025

Frontier models design the agent. Smaller models (30B-120B) do the work. Deterministic tools guarantee consistency. Bigger isn't better. Smarter is. 🔗 dev.to/dexmac221/the-giant-that-builds-smaller-giants-custom-ai-agents-for-privacy-and-efficiency-1ed0 #AI #LLM #Ollama #Privacy #MoE

dev.to

The Giant That Builds Smaller Giants: Custom AI Agents for Privacy and Efficiency

The future of AI isn't bigger models. It's smaller, specialized agents — distilled, custom-built, and...

0132

Dexmac @dexmac.bsky.social · 23/12/2025

Merry 8-bit Christmas! 🎄 Snow falling on a C64. Demo generated with my AI toolchain — modern AI, vintage hardware. github.com/dexmac221/C6... #commodore64 #c64 #christmas #retrogaming #AIgames #generativeAI

0246