Home-lab AI · localdeploy.dev

Turn idle hardware into an assistant you use every day

Assess the hardware first, read the free basics, then choose a paid deployment package if you want it done. No fear-selling, no benchmark promises: every step written down.

Free assessment · no sign-up 中文 / English Packages quoted by scope
homelab@localdeploy
$ free -h
               total        used        free
Mem:           31Gi        9.2Gi        18Gi

$ nvidia-smi --query-gpu=name,memory.total --format=csv
name, memory.total [MiB]
NVIDIA GeForce RTX 3060, 12288 MiB

$ df -h /srv
/dev/nvme0n1p2  1.8T  612G  1.1T  36% /srv

# 32 GB RAM + 12 GB VRAM: the comfortable range for 14B Q4
3free long-form guides
6deployment areas
1standard assessment text
0tracking scripts
Resource assessment

What can this machine actually run?

Fill in what you already own. The result is a rough starting point for a conversation, not a benchmark: all of it is calculated in your browser and never uploaded.

Hardware

Nothing is sent by this form. The email button only opens your mail app with the text filled in.

Result
Fill in the fields and press "Generate assessment".

The result is calculated locally; nothing is uploaded.
It will list: RAM, disk, GPU/VRAM, upstream bandwidth,
a rough model tier, and what to check next.

The assessment stays in this tab. Email is only sent when you choose to send it.

Free basics

Start with the three things that decide the rest

No account, no email gate. Read them in order or jump to the one that matches your current problem.

Paid packages

When you want it running, not just explained

Six areas, one rule: scope is written down before work starts. Prices are quoted after a short scope check, because a Proxmox box with one container and a three-node cluster are not the same job.

Current package

A fixed-scope review you can act on

Not a course and not an open-ended engagement. One review, one written plan, and a list of what to change first.

Planned v2026-11

Home AI lab deployment review

A remote review of your hardware, system and services, with a written plan: what can run, where the bottleneck is, and what to change first.

Quote on request View scope

No online checkout is live at this stage. Read the scope, then send your setup by email.

How it goes

From a hardware snapshot to a working box

The assessment is free and local. Everything after that is scoped in writing before any work starts.

  1. Measure first

    Use the assessment to turn CPU, RAM, disk, GPU and bandwidth into one standard text.

  2. Fill the gaps

    Read the Cloudflare, security and system guides to decide whether the limit is hardware, network or the operating system.

  3. Pay only if needed

    Send the assessment and your goal by email. Scope and a written quote follow; if it is not a fit, the reply says so.

  4. Leave with something you can maintain

    Configuration, documentation and verification steps, so the next change is yours to make.

Why this site

Why the free material is this detailed

Most problems are not missing an advanced trick; they are basic concepts that never lined up.

01

Decide before buying

Know what your hardware and network can do before you buy or install anything.

02

Write the boundaries

Free material makes no guarantees; paid work does not take over your accounts or run 24/7.

03

Readable by agents

Every page has a Markdown mirror and structured metadata, so your own assistant can read and cite it.

Boundaries

What this site does not do

Saying what is out of scope is more useful than saying how professional we are.

01

No hardware sales

No shop links and no affiliate commission; only the dimensions that matter.

02

No benchmark promises

The same model varies with drivers, quantization and context length; you get a tier and a way to verify it.

03

No account handover

Paid work happens on your machine; your passwords and tokens are not held long term.

Questions

Answers before you send an email

I have never used Linux. Can I follow this?

The concepts and the order of decisions are readable, but practise on a machine you can break. The free material does not assume you write code.

Is the assessment accurate?

Not to benchmark level. It uses VRAM, RAM, disk and bandwidth to place you in a rough tier, so you can decide whether to try and what to check first.

Why not sell a course?

A course struggles to cover every home setup. This site separates the method (free) from the work in your environment (paid, fixed scope).

Can a paid package guarantee it will work?

No. What can be verified in your environment is verified; third-party, driver and model limits are outside that. Scope is confirmed before any quote.

Is there a Chinese version?

Yes. Use the language switch in the header; the main pages have a Chinese counterpart.

Do you collect my hardware details?

No. The assessment runs locally in the browser, with no upload, no cookie and no analytics. Information leaves the browser only if you send an email.

Who this site is for

  • People with an old PC, NAS, mini PC or workstation who want to run a local model but are not sure the hardware is enough.
  • People already using Docker or a NAS who want a maintainable structure: one entry point, private images, monitoring and backups.
  • Small teams that want shared in-house AI without sending data to an external API, and without buying a server on day one.
  • People willing to read, test and accept “assess first, then build”.

Who it is not for

  • Anyone looking for a one-click, zero-maintenance setup. Drivers, quantization and runtimes change too often for that promise.
  • Anyone looking for a guaranteed performance number. The same model varies with the machine and context length.
  • Anyone who wants an account handed over long term. Paid work happens on your machine and is handed back to you.

Why start with the assessment

The limit in local AI is rarely “the CPU is not fast enough”. It is usually VRAM, RAM, disk or network, whichever runs out first. Writing the configuration down as a standard text makes both self-research and an email exchange far more useful.

The assessment gives an order-of-magnitude tier: not tokens per second, but whether your machine is in the light, starter, balanced or advanced range, and whether RAM, disk, multi-GPU or bandwidth deserves attention first.

How the content is maintained

Free pages are maintained by topic, each with its own update date. Paid packages are quoted per project, with scope written into the quote. Dates come from frontmatter rather than being copied by hand. If you find an error, send an email; the address is in the footer.

Next step

  1. Run the assessment and copy the result.
  2. Read the one free guide that matches your current blocker.
  3. If you want a working result, email the assessment and your goal to hello@localdeploy.dev.
Reading this as an agent? Every page has a Markdown mirror: /index.md, /cloudflare.md, and one per page. The index is /llms.txt, the capability table is /skill.md, discovery metadata is /.well-known/ard.json. Send Accept: text/markdown to a page URL to get the Markdown mirror.

Not sure where to start?

Run the assessment first, or send your current setup and goal by email. The reply says whether the scope is a fit and what is needed next.