FAQ

The download is not the climb.

Anyone can fetch a model. We learn your profile, point you at the newest one that fits this computer, and stay for the workflows — how to use it, how to teach it, how to point it at the job.

Is it actually private?

Yes, by design. After the first download, the model runs on your computer. What you type is not sent to us, or to OpenAI, or to Anthropic. There is no server in the middle that could read it.

Can you see my conversations?

No. We cannot. The weights live on your machine. We interview you about your job and your computer so we can write a plan. We do not see the files, notes, or chats that follow.

Is this just another download button?

No. Downloading a model is the first mile. The work is knowing which model fits you this month, how to point it at your job, and how to build the workflows after it is on the disk. That is the Sherpa. A dashboard that only fetches weights leaves you at the trailhead.

How do you pick the newest models?

Every Monday we re-read Hugging Face and re-rank the top nine in each list — by country, and by whether it runs on your computer or needs a private server. The Field Guide is that list, in ordinary language. Then we match one to your machine. A dentist on 16 GB does not get the same checkpoint as a CFO on a private GPU box.

Why do you need my role?

Because a nurse, a CEO, and a student are not the same climb. Jenna asks eight questions so the plan, the model, and the workflows match the work — not a generic “assistant.” We understand the profile. Then we point.

What happens after it is installed?

We stay for the part most installers skip: how you use it. Letters, lesson plans, board notes, a local file vault, a first training run on examples you choose. The local guide remains on the machine after we leave. You own the summit.

Is this a competitor to ChatGPT or Claude?

No — and that is on purpose. Those tools run on someone else’s computer. They are the right call for public, throwaway work. If the material cannot leave the room, they are the wrong computer. We install an open-source model on yours, then teach you how to point it. If you are already fine in ChatGPT, we send you to Connect, not a sealed install you do not need.

Does it work without internet?

Yes — after the first download. On a plane, in a clinic, with Wi-Fi off. Try it: unplug mid-conversation. It keeps going, because it is running on your computer, not in a data center.

Why is the first download large?

You are bringing the model onto the disk once, instead of borrowing a brain over the internet every time you ask. Like downloading a film for a flight. After that it is local, and it stays.

What do I need to run it?

A computer you control and enough memory for an honest model. 16 GB is a real start. 8 GB will run a small one. We check the machine before you pay. If it cannot run, you are not charged. If a model is not answering in 14 days, you do not pay.

Why isn’t this free?

The model is open source. The walk is the work. We charge our cost plus 20%. That 20% pays the people who accompany you. We do not take more. Gifts and sponsored seats go to the Summit Fund, a 501(c)(3) in formation — not tax-deductible until it can issue a receipt.

Can I use ChatGPT for some of it?

If the material is already public, or you invented it for practice — yes. That is Connect. If it might be a student record, a chart, or a client file, we stay offline. When we are not sure, we stay offline.