Architecture
Choose boundaries between routing, storage, compute, management, and user-facing services.
FG / 01 / Field guide
Home Lab
Build the quiet, recoverable backbone for local AI, storage, networking, and the services worth owning.
Pulse
Current briefing
What is moving now.
Standing watch
Compact 2.5GbE switching
Serve the Home
Watch for quiet, low-power options that improve the next rack revision without creating a management or compatibility tax.
Open the brief →Primary-source watch
Proxmox platform changes
Proxmox documentation
Track dated release notes that materially affect backup, clustering, storage, recovery, or hardware support.
Open the brief →Build watch
Low-power NAS tradeoffs
Hardware Haven + NAS Compares
Collect practical builds that disclose power, noise, thermals, expansion limits, software choices, and failure modes.
Open the brief →Current briefing
FCC Covered List Expansion Targets Foreign Consumer Routers
Protectli
An FCC Covered List update blocks regulatory authorization for new foreign-made consumer routers, raising boundary questions for modular edge hardware and home lab firewall appliances.
Open the brief →Current briefing
License-Free GPU Slicing Lands on Proxmox via Intel Arc Pro B-Series
Level 1 Techs
Intel Arc Pro B60 and B70 GPUs enable subscription-free SR-IOV virtualization on Proxmox VE, offering home labs and small enterprises a low-cost alternative to recurring Nvidia vGPU licensing.
Open the brief →01
Get your bearings
Treat the lab as a system with five layers: network, compute, storage, applications, and operations. The rack is not the product. Reliable services, controlled failure, and understandable recovery are the product.
02
Start here / Three moves
- 01
Map what already exists
Inventory hosts, storage pools, network dependencies, power, backups, remote access, and the services that other people depend on.
- 02
Design the NAS migration
Write the target architecture, data classes, backup path, validation checks, cutover order, rollback plan, and acceptable downtime before moving anything.
- 03
Build an operating baseline
Record idle power, noise, temperatures, update cadence, alerts, credential handling, and one tested restore.
03
Build capability
Operations
Patch, observe, back up, restore, document, and retire systems without depending on memory.
Local AI
Match models and accelerators to real workloads, then account for thermals, power, memory, and data sensitivity.
04
Choose gear / Decision rule
Choose by workload, consequence of failure, idle power, noise, expansion, software compatibility, recoverability, and full-system cost—in that order. A spec sheet without an operating plan is not a recommendation.
before
affiliate
05
Your source map
Start with voices you already value. Verify before concluding.
YouTube
9App / Tool
2- ProxmoxApp / Tool
- OPNsenseApp / Tool
Brand
4- ProtectliBrand
- NetgearBrand
- TP-LinkBrand
- UGREENBrand
Reference layer
Use these to settle claims—not to manufacture certainty.
Field Notes
Original editorial work
Independent analysis built from the strongest available evidence.
Field NotePublished
The rack is not the product
A useful home lab is measured by recoverability and capability—not the amount of equipment in it.
Read the working note →Project BriefNext brief
A NAS migration that does not become a dependency trap
Architecture, validation, rollback, and the operating baseline that should exist before the first file moves.
Read the working note →Field NotePublished
The Opinionated Linux Desktop: Evaluating Omarchy and the Turnkey Tiling Stack
Pre-configured Arch Linux distributions promise the speed and aesthetics of custom tiling window managers without the hundreds of hours spent editing configuration files. Here is what they actually deliver—and where the friction remains.
Read the working note →Field NotePublished
The Two Sides of VPNs: Privacy Tunnels vs. Private Gateways
Commercial VPN ads sell fear, but the real utility is twofold: routing outbound metadata away from your ISP and safely tunneling back into your home infrastructure without exposing insecure ports.
Read the working note →Field NotePublished
The 48GB RTX 4090 Mod: Board Transplants and Local AI Economics
Doubling Ada Lovelace VRAM requires harvesting the core onto a custom clamshell PCB. For self-hosted AI workloads, the hardware delivers workstation memory density—at the cost of warranties and thermal compromises.
Read the working note →Field NotePublished
The Scrappy Homelabber’s Guide to Sourcing Used Enterprise and PC Hardware
Secondary market prices for used computers are inflated, but systematic sourcing strategies—from local ITAD pipelines to asymmetric search queries—can still net capable homelab silicon at scrap prices.
Read the working note →Field NotePublished
Opinionated Arch Desktops and the Daily Driver Dilemma
Pre-configured tiling environments promise to eliminate Linux configuration fatigue, but curated defaults cannot fully bypass the realities of rolling-release workstation maintenance.
Read the working note →Field NotePublished
Beyond the AI PC Hype: Assessing Integrated NPUs for 24/7 Homelab Inference
Dedicated neural accelerators are finally arriving in compact homelab hardware, but their true utility hinges on runtime stability and pipeline offloading rather than raw throughput.
Read the working note →Local AI in Practice
Capabilities / Workflows
Start with the jobs local AI handles well, then combine those capabilities into useful workflows at the 16, 32, and 64 GB classes.
Start here
What local AI is actually good for
Use local AI for frequent, private, narrow, and reversible work. Use frontier models for difficult reasoning, consequential decisions, current research, and final output where the highest available quality matters.
Hardware shorthand
Usable AI memory—not the sticker on the box.
Fast local utilities: transcription, OCR, extraction, classification, short summaries, voice intent, and focused assistance.
Stronger models, longer context, private document retrieval, more natural interaction, and dependable multi-step workflows.
Larger models and knowledge bases, richer multimodal work, and several local stages combined into a serious private workflow.
Capability families
Combine these building blocks into workflows.
Transcribe speech, read images and screenshots, OCR documents, and recognize useful events or structure.
Summarize, extract, clean, classify, tag, rename, redact, and route material.
Search private documents, manuals, notes, logs, code, and other defined knowledge.
Draft text, code, commands, configurations, explanations, and first-pass output for review.
Watch bounded inputs, identify changes, trigger approved actions, and hand work to another tool, model, or person.
Value can look like
- Time or attention saved
- Privacy and local control
- Lower cloud tokens or cost
- Lower latency or offline resilience
- Quality-of-life improvement
- Learning, creativity, family use, or simple fun
Do not force local AI into
- Consequential legal, medical, financial, or security conclusions
- Tasks whose answer depends on current outside information
- Deep open-ended reasoning better handled by a frontier model
- Unreviewed destructive administration or unrestricted tool authority
- Final work where frontier-level quality matters more than privacy or speed
- Elaborate setups whose maintenance costs more than the result is worth
Use CaseProven pattern
Frontier-model preparation layer
Let local AI transcribe, extract, redact, organize, and compress the source material before the difficult reasoning goes to a frontier model.
Open use case →Use CaseProven pattern
Home Assistant voice control
Use local speech and language models as a private, resilient interface to approved Home Assistant actions—not as an unrestricted administrator.
Open use case →Use CaseProven pattern
Automatic document and file organization
Turn designated inbox folders into a reviewable queue of extracted metadata, sensible filenames, destinations, summaries, and duplicate warnings.
Open use case →Use CaseProven pattern
Autonomous Local Coding: Self-Hosted Orchestration with Ollama, OpenHands, and Smart Routing
Replace recurring cloud API token bills and keep proprietary codebases private by pairing local Ollama runtimes with OpenHands agent workflows and smart routing proxies.
Open use case →Use CaseProven pattern
Self-Hosted Model Gateway: Local LLM Routing and Task Execution
Cut API costs and eliminate latency by routing routine coding, summarization, and tool tasks through homelab hardware while preserving frontier cloud APIs for heavy reasoning.
Open use case →Gear
Recommendation ledger
What fits, who it fits, and where it fails.
recommendedOwned
Protectli Vault
Firewall applianceCandidate
Open the product record →StandardEvery approved pick
Who it is for. Why it fits. Where it fails.
Each recommendation will include drawbacks, alternatives, last-reviewed date, evidence level, and affiliate disclosure.