The same prompt is not a repeatable answer
A model can answer the same question differently on two runs, even at temperature zero. Treat one response as a sample, record what produced it, and check the spread rather than trusting a single reply.
Articles
38 published · same source policy as the coursesShort, reviewed pieces on practical AI topics. Every factual claim is registered against a primary source, the same honest-claims and source policy the course lessons follow — these are not marketing copy or unreviewed drafts.
A .env file is not secret management
A .env file keeps a value beside the process. Secret management still has to answer where that value lives, who can read it, how it is replaced, and how the backup is protected.
A restart policy is not supervision
A restart policy brings a container back after it exits. Supervision is the separate work of noticing the repeats, limiting them, and escalating while the process is still failing.
A UPS buys minutes, not a clean shutdown
A UPS keeps the power supply alive for a while. The server still has to hear about the outage, shut down inside the runtime it actually has, and know what "power is back" is supposed to look like.
A passed SMART test is not a healthy disk
A drive that reports PASSED has cleared one vendor check. Record a baseline, schedule self-tests, watch a few raw counters for change over time, and have a restore you trust before you act on any of it.
A model card measures someone else's task
A model card and a public benchmark report someone else's test. Freeze a small set of your own cases, name the failure you refuse, and re-run that set before a model does the job unattended.
When a person moves the switch, the automation should yield
A home automation that re-asserts its own state after you touch a switch is not being helpful; it is arguing. Use the cause recorded with every state change to decide who owns the final say.
A workflow is not an agent
Calling a fixed sequence of steps an "agent" hides the design decision that matters. Before you hand control flow to a model, check whether the path is one you can draw — and start with the simplest structure that works.
Retrieval is not verification
Adding a search step to an AI assistant makes answers look grounded. Retrieval decides what text the model sees; it does not check that the text is right, current, or actually supports the claim.
Where VLAN isolation actually comes from
A VLAN separates traffic at layer two, but the separation lives in every link on the path — and in the discovery services you choose to re-bridge. A practical way to see where segmentation is real and where it has quietly disappeared.
What a rollback can and cannot undo
A rollback command returns your code to an earlier revision, not the world around it. A practical way to separate the artifact you can undo from the schema, data and side effects you cannot, and to design a release so the answer is honest.
Before your automation retries, make the action idempotent
Retries turn a brief failure into a duplicate side effect unless the action can be repeated safely. A practical checklist for stable operation identifiers, precondition checks, outcome records and bounded retries, and an honest account of what it cannot promise.
A missing sensor reading is not zero
Unknown, unavailable and stale sensor values are not measurements. A practical way to keep missing data out of averages, thresholds and automations, and to leave a household with a safe manual path.
What you saw, what you think, what you changed
In an AI-assisted debugging session, a guess about the cause is quickly repeated back as a fact. Keep observations, inferences and actions in three separate records, and let a claim move between them only on new evidence.
Approval should bind to the change, not the conversation
A "yes" given in a chat drifts away from the thing it approved. Tie each approval to one exact, identifiable version of a change, and treat any later edit as needing a fresh decision.
When your home server clock is wrong, evidence stops lining up
A wrong clock can scramble the apparent order of logs, move calendar jobs and make certificate checks fail. Diagnose the time source and offset before correcting it, then verify the evidence again.
A web page is not permission for your AI agent
An agent can use a page, document or message as evidence without treating instructions inside it as authority. Design that boundary before connecting powerful tools.
Automate disk cleanup only after you can explain the growth
A full disk invites a hurried deletion. A safer home-server workflow identifies the owner, growth rate, reclaimable data, retention need and recovery path before cleanup becomes automatic.
Local AI stays private only if the whole workflow does
Running a model on your own machine can remove one remote data path, but prompts, tools, logs, saved outputs, backups and sharing still decide where information travels.
Draft first, send second
AI can help shape an outbound message without deciding who receives it or pressing send. Use a bounded draft, a deliberate human check and a clear correction path.
What your smart home does when the internet is gone
Map every smart-home dependency before an outage: where decisions happen, how updates arrive, what fails and which safe manual path remains available.
A health check is not a recovery plan
A practical home-server guide to separating liveness, readiness and restart signals from the human decisions that make recovery safe.
Cloudflare Tunnel is not a trust boundary
A tunnel can keep an origin off the public internet, but it does not answer who should use the application. Separate routing, identity and application permissions with a small, testable checklist.
Give your AI agent a budget
Permissions limit what an agent may touch. A separate budget for time, steps, data and queued work limits how much it can consume before it must stop.
An agent's memory is not an audit trail
Memory helps an agent continue a task, but it cannot prove what happened. Separate useful context from a small, reviewable record of each run.
The log says done. Prove it.
A practical evidence trail for AI-assisted work: pin the inputs and revision, record what actually ran, separate review from authorship, and keep a recovery path.
A firewall rule is not a network boundary
A practical OPNsense-inspired checklist for separating trust zones, writing understandable rules, testing the path and keeping a recovery route when a firewall change goes wrong.
Don't ask AI if it's right
A confident answer is not evidence. Build a small verification plan around the claim, the consequence of being wrong, and the source or test that can actually settle it.
A scheduler is not an owner
A practical way to design scheduled automation around missed runs, duplicates, evidence, approval and recovery instead of assuming a clock makes work reliable.
The alert you always ignore
An alert nobody acts on is not monitoring. It is noise with a notification sound. A practical way to decide which home-server alerts deserve to interrupt you, using precision, recall, detection time and reset time.
What AdGuard Home can and cannot fix
What network-wide DNS filtering genuinely solves, the ceiling AdGuard's own documentation sets out, and how encrypted DNS and Encrypted Client Hello change the picture.
Keep your AI agent on a short leash
Why the real risk of an AI agent is the authority you hand it, not the model going rogue — and a practical way to turn down its functionality, permissions and autonomy separately.
Make home automations fail safe first
A practical checklist for designing Home Assistant automations that recover clearly, leave evidence and avoid doing the wrong thing twice.
Home Assistant automations that fail clearly
A practical way to design household automations so missed conditions, failed actions and recovery steps are visible instead of mysterious.
A backup is not a restore plan
A practical home-server guide to turning backup jobs into evidence you can actually recover from, using restore drills, integrity checks and clear limits.
Before you paste that into an AI chat
A practical data-confidentiality guide for anyone about to paste a document, email or dataset into a general AI chat tool — what actually counts as sensitive, what happens to it next, and where the UK's data protection guidance sits.
Choosing your first local model
A five-minute, practical way to narrow down your first local AI model — and where to go for the deeper, hands-on version in the Local AI Agents course.
What quantisation actually does to your model
A standalone explainer of what "quantisation" means for a downloadable AI model file, what it actually costs, and how to read a name like Q4_K_M before you download it.