Know what is included.
Start with one checked device and a defined task. Agree the scope and acceptance sample before committing.
INDEPENDENT HERMES SETUP & GUIDANCE
Get an AI agent installed for real work.
Get set up, test it and review it before you pay.
Our upcoming promise: request a full refund of the LocalVRAM service fee within 72 hours of payment, no reason required. Future-service draft policy; service and refund channels are not live.
- Remote setup: get Hermes installed on a compatible device.
- Configuration: connect the model included in your package.
- First working task: put your setup to use on an agreed everyday task.
- Guided handoff: practice the steps so you can run the task again.
- Scoped aftercare: get help with the agreed configuration during your package’s support window.
A GUIDED START
Start with help. Leave with a task you can repeat.
Learn it yourself.
Practice the agreed task and review the output. Your handoff includes setup boundaries and how to stop unneeded access.
Keep account control.
You enter credentials directly. Remote work needs explicit permission for the device, actions and time window.
SERVICE OPTIONS
Start with the help you need.
Choose the level of setup, practice and aftercare that fits your work. Each package includes a guided introduction and a defined configuration support window.
A FIRST EVERYDAY TASK
Remote Hermes Setup
USD 99
- 1 compatible device
- 1 model configured
- 30-minute hands-on introduction
- 7 days of agreed configuration support
A first working setup and everyday task.
ONE LIGHT PERSONAL WORKFLOW
Personal AI Onboarding
USD 299
- 1 compatible device
- Up to 2 models configured
- 1 light personal workflow
- 60-minute hands-on introduction
- 14 days of agreed configuration support
One workflow you can repeat and review.
ONE AGREED TEAM PROCESS
Small Business AI Setup
FROM USD 999
- Up to 3 compatible devices
- Up to 2 models configured
- 1 agreed business process
- Up to 120 minutes of training
- 30 days of agreed configuration support
Model allocation across devices is agreed in writing.
Direct third-party API, model, subscription and provider charges billed independently of LocalVRAM service fees are separate costs, disclosed upfront. Only those independently billed charges may be excluded from the service-fee refund, with upfront disclosure. Custom development, unlimited support, dedicated GPUs and cloud hosting require a separate quote.
Support covers the agreed configuration during the stated window, starting at written acceptance. It excludes new workflows, provider outages, new devices and third-party changes. Contact channel, hours and response target must be agreed in the quote; no 24/7 or unlimited support is included.
DEVICE FIT COMES FIRST
A specific route. A transparent precheck.
Windows 11 + WSL2
Existing reviewed WSL2 routes require a customer-specific precheck and manual quote. Some first-time WSL2 cold installs remain unaccepted. No universal native Windows/MSIX or instant-install promise.
Apple Silicon Mac M4
Apple Silicon Mac M4 installation demonstrated. Each Mac requires a device-specific fit check. Installer signing/notarization and Intel Mac compatibility require separate validation. This demonstration does not establish support for every Mac. Mac GUI and other OS routes need separate review.
Advanced work: separate review
Browser automation and Multi-AI Workspace remain pilots. GPU optimization, payment engineering and editorial solutions require separate validation. Scope and readiness are assessed separately.
Conditional add-ons need device and scope review; extra work must not duplicate anything included in your base package.
HOW SETUP WORKS
See it work before you pay
Check fit and agree scope
Review your exact device route and agree a nonsensitive task, basic acceptance criteria, costs, support and terms before the session.
Set up remotely
Authorize a bounded remote AI Agent setup session and enter your own credentials.
Test and review
Run a basic functional test, review the result and record your basic acceptance. Review the handoff and revoke unneeded remote access.
Pay after basic acceptance
Under the future-service draft, payment is requested only after that basic acceptance. Once paid, you have 72 hours from payment to request a full refund of the LocalVRAM service fee, no reason required.
BEFORE YOU COMMIT
Clear answers, upfront.
72-Hour Money-Back Guarantee
Under this future-service draft policy, remote AI Agent setup comes first, followed by a basic functional test and customer review. Payment is requested only after that basic acceptance. Within 72 hours of payment, you may request a full refund of the LocalVRAM service fee, no reason required. The 72-hour request window starts when payment is made, not at setup or acceptance. No deduction is made from that service fee for work already performed.
Direct third-party API, model, subscription and provider charges billed independently of LocalVRAM service fees are excluded only with upfront disclosure. Applicable statutory rights remain in addition. Service and refund channels are not live; see the draft refund policy and future request method.
What do the catalog prices cover if Hermes is open source?
The proposed packages cover independent engineer time for installation, configuration, teaching and scoped support. Hermes Agent is open-source software from NousResearch. LocalVRAM is independent of NousResearch, is not an official partner and claims no endorsement. Hermes Agent remains the upstream project; these packages describe independent services.
Will this work on my Windows PC?
The initial route is Windows 11 with an existing reviewed WSL2 environment, followed by a customer-specific precheck and manual quote. Some first-time WSL2 cold installs remain unaccepted. There is no universal native Windows/MSIX or instant-install promise.
What about Mac or another operating system?
A real Apple Silicon Mac M4 installation has been completed. We check each Mac’s device and setup route before agreeing scope. That result does not confirm every Mac, installer signing/notarization or Intel Mac compatibility. Mac GUI and other operating systems need separate review.
Does LocalVRAM mean everything runs privately on my GPU?
No. The selected model and tools determine where processing happens. A hosted model may send prompts to its provider. Local-only inference and GPU optimization are separate validation topics, not included promises.
How are credentials handled during setup?
You enter your own credentials directly and authorize the device, actions and time window for remote work. Use a nonsensitive sample for the first task and revoke unneeded access at handoff.
How long does setup take?
The package specifies introduction or training time, not a guaranteed installation deadline. Device checks, dependencies and scope determine scheduling. Availability is agreed before a session.
What happens if my device or task is unsuitable?
Pause and review the scope. A different scope needs a new agreement. Under the future-service draft, payment is requested only after setup, a basic functional test and customer review result in basic acceptance. Final cancellation and rescheduling terms still need review before service opens.
What support is included?
Seven, fourteen or thirty days of agreed configuration support, depending on the base package, starting at written acceptance. Hours, channel and response target are written into the quote. New devices, new workflows, third-party changes and provider outages are outside that boundary.
ONE CLEAR STARTING POINT
Find the right start for your first working task.
Compare device counts, model configuration, guided practice and aftercare in one place.
Compare packages