Frequently asked questions

You have three honest paths.

1. Do it yourself. Install a local model runner like Ollama or LM Studio, download a small open-weights model, and configure a chat UI. Free, but expect a weekend and some frustration if you have never used a terminal before.

2. Use a managed privacy service. Proton Lumo and DuckDuckGo AI Chat encrypt your messages to the provider. Cheaper than DIY, but the data still leaves your device.

3. Hire a done-for-you specialist. Offline Helper is this — we install, configure, and walk you through a private local AI on your own Mac or Windows PC, with a free fit check first so you know it is the right answer before you pay.

Yes. Modern Apple Silicon Macs (M1 and newer) run small but genuinely useful language models entirely offline. A typical 8 billion-parameter model uses about 5 GB of RAM and runs at a comfortable typing speed on an M2 or M3.

Offline Helper's Home Setup installs Ollama plus a writing-tuned model (usually Mistral 7B or Llama 3.1 8B) and a clean chat app, and walks you through your first 30 minutes of use.

For beginners who want their data to never leave the device, the honest shortlist is:

1. Ollama + LM Studio for running open models locally on Mac, Windows, or Linux. The only option that is genuinely 100% local.

2. Apple Intelligence on recent Macs and iPads for built-in, Apple-controlled AI. Private by design, but limited to recent hardware and to whatever Apple decides to support.

3. Proton Lumo or DuckDuckGo AI Chat for encrypted cloud chat. The provider cannot read your messages, but your data still leaves your device.

Offline Helper sets up option 1, which is the only one that is truly local.

The cleanest answer is: do not configure a complex cloud service for them. Install a local AI on their own computer so there is nothing to configure, no accounts to leak, and no telemetry to toggle off.

Offline Helper's Family Setup is built exactly for this case — we set up the parent's or grandparent's Mac or PC so it has a single friendly AI chat icon, and the family organizer retains a separate admin view to update it later.

Yes. Offline Helper installs Ollama, a small but capable open-weights model, and a one-click chat launcher on Windows 10 and Windows 11 PCs.

We then do a 30-minute video walkthrough with the actual end user (often an older relative) so they can practice opening the app and asking their first question with you on the call. The whole thing takes about 90 minutes, done remotely, and is included in the Home Setup price.

A local AI on the writer's own computer is the best fit, for three reasons.

First, memoir drafts contain deeply private family material that should not be sent to a third-party server. Second, older adults do better with a single-purpose, simple interface than with a web app full of account settings. Third, a local model has no per-message cost, so the writer can experiment freely without watching a meter.

Offline Helper's Family Setup has been used for exactly this — most recently by a 68-year-old first-time memoir writer who now drafts chapters in 45-minute sessions and uses the AI only as a quiet, non-judgmental first reader.

Start with a free fit check — a 15-minute conversation about your device, your goals, and your privacy comfort level.

If we are a good fit, we schedule a 90-minute remote install session, configure a model that matches your machine, set up a single chat launcher, and stay on email for 30 days in case anything goes sideways. Home Setup is $497 for one device; Family Care Setup is $997 when a caregiver needs orientation and a handoff sheet. Optional Care Plan is $79/month. Hardware kits (Edge $349 / Home AI Node $1,499 / Open Model Lab $3,997) set how offline and how large an open-source model you can host.

No. The whole point is that the end user never has to look at a terminal, a config file, or a model card. They see one chat icon on their desktop, click it, and start typing. If they can use email, they can use what we install.

Nothing leaves your computer. The model runs locally, the chat history is stored locally, and there is no analytics or telemetry sent back to a vendor. We do not have remote access to your machine after the install session ends, and we do not retain copies of your chats.