Blog.
Technical insights on AI compute, memory orchestration and deployment where the cloud cannot reach.
[ Latest ]
How to build offline RAG that works air-gapped
Build offline RAG for an air-gapped site: serve an embedding model with llama.cpp, store vectors in SQLite and log what every answer retrieved.
How to run vLLM and llama.cpp fully offline
Run vLLM and llama.cpp with no internet: stage weights, wheels and images, set offline env vars, fix errors that break air-gapped installs and test it.
Google's London outage came down to one PLC check
Google's incident report on the 20 October europe-west2 outage: generators never started because a PLC read voltage on the wrong side of an open breaker. What that means for anyone running AI on critical sites.
[ All Posts ]
RSSHow to build offline RAG that works air-gapped
Build offline RAG for an air-gapped site: serve an embedding model with llama.cpp, store vectors in SQLite and log what every answer retrieved.
How to run vLLM and llama.cpp fully offline
Run vLLM and llama.cpp with no internet: stage weights, wheels and images, set offline env vars, fix errors that break air-gapped installs and test it.
Google's London outage came down to one PLC check
Google's incident report on the 20 October europe-west2 outage: generators never started because a PLC read voltage on the wrong side of an open breaker. What that means for anyone running AI on critical sites.
Running a model locally is not controlling it
Cohere caught an open-weight model trying to upload a file to a cloud bucket during a sandboxed test, while its reasoning said it would use no tools. Local is not the same as under control.
Open weights are the new strategic reserve
Open-weight models are AI's strategic reserve, but total parameters set the memory bill. The math for running models like Mistral Large 4 with no cloud.
The data centre just moved under the desk
DGX Spark 64GB goes on sale 23 Oct 2026 at US$4,999. Local AI compute is now a desk item. Running it safely at every site is the harder problem.
Google Cloud London outage: zones are not a continuity plan
A power event in one Google Cloud London zone on 20 Oct 2026 showed up as errors across the region. Zone redundancy is a design pattern, not a continuity plan.
AI in space: orbit is the hardest edge
Google's Suncatcher put four TPUs in orbit on 1 October 2026. AI in space is the edge with the excuses removed: fixed power, no repairs, silent bit flips.
Your AI has an off switch. Who holds it?
A June 2026 AI export control letter switched off two frontier models worldwide. Critical AI needs a path that survives someone else's policy change.
The cloud has a blast radius: AI continuity in 2026
War, heat, lightning, an automated account flag and ransomware took cloud regions down in 2026. AI in critical systems needs a smaller blast radius.
Self-replicating AI will live on compute nobody owns
Anthropic says small AI models may soon self-replicate across cheap IoT devices. The risk is not edge AI. It is edge compute with no operator in command.
Data centres are now targets
Drone strikes on data centres in Russia and Ukraine show a cloud region is a building. AI in critical systems must keep running when one goes dark.
Location is not control: Australia's AI sovereignty test
Megawatts and postcodes are not sovereignty. Australia's AI Standards must force a visible control chain for Commonwealth and critical infrastructure AI, or the country will certify racks while someone else owns the layer that matters.
Who can stop the model? Operator control for AI at remote sites
Operator control for AI at remote sites: who can stop a model, approve a change and roll it back, and what audit trail proves it, even with the link down.
GPU memory planning: a guide to VRAM and system RAM
A GPU memory planning guide for AI inference: budget VRAM for weights, KV cache and overhead, then plan system RAM as the second tier before you buy.
Running AI on intermittent links: a design guide
How to run AI at sites with satellite, cellular or radio links that drop: local inference, store and forward, thin-link model updates and clock discipline.
How to fix vLLM OOM (CUDA out of memory): 8 steps
vLLM crashing with CUDA out of memory? Fix it in 8 steps, from gpu_memory_utilization and max_model_len to quantisation and memory tiering.
What is sovereign AI? Definition and 5 tests
Sovereign AI is AI you fully control: the model, the data and the infrastructure stay under your legal and physical control. Definition, examples and 5 tests.
The On-Premise AI Gap: Why Cloud Isn't Always the Answer
Cloud AI APIs are convenient, but they're not viable for defense, government, healthcare, and industrial operations. Here's why on-premise AI infrastructure matters.
Why We Built Sector88
From solving GPU memory crashes in production ML systems to building infrastructure that makes on-premise AI deployment actually work. The story behind Sector88.
Introducing Sector88: AI compute for constrained hardware
Running AI should not require unlimited GPU budgets or a cloud connection. How Sector88 runs AI compute on constrained hardware, starting with the memory wall.