60 minutes
Executive AI demo showcase
Live demos of AI running locally, built for the customer's industry and branded for their company. It opens a conversation that goes beyond the PC refresh.
What we deliver
Everything your team needs to sell local AI, from the first meeting to a working proof of concept.
Working session
AI workload planning session
We review the customer's work and data, decide where each AI workload should run, and pick one to prove. You leave with a defined prototype and a clear measure of success.
Built for one customer
Customer-branded prototype
A working demo of the customer's own AI workload, in their brand. It runs on the device with the network off.
Two four-hour sessions
Windows Local AI Bootcamp
Sales teams: a local AI 101. Speak the language, spot the customers ready to adopt and know who decides on AI workloads.
Architects and technical sellers: set up models, build demos and test workloads on Windows.
Each track includes Level 1 certification.
On call
On-device AI help desk
Your sellers can reach a solution architect when a customer asks about workloads, models or hardware. A technical question shouldn't stall a deal.
30 days
RTX Spark Accelerator
We take one customer AI workload from plan to working proof of concept on Surface Laptop Ultra. We handle the technical work with your team. You own the customer relationship. See how it works.
Ongoing
Sales and marketing support
Event speakers, co-branded webinars, competitor guides, talking points and demo kits. Plus a co-branded planning tool that helps customers compare costs and decide where each AI workload runs.
Why now
The first wave of AI brought your data to the model. The next wave brings the model to your data.
Customers are blocked
- Data that can't leave the building
- Rising AI bills
- Pressure to move faster
The hardware is here
New Windows devices built to run AI locally:
- Surface Laptop Ultra
- RTX Spark laptops and desktops
- Surface RTX Spark Dev Box
- DGX Station for Windows
The gap remains
Customers still need answers:
- What should run where?
- How do we govern the agents?
- How do our teams build on it?
Partners need help building a local AI practice.
Customers need an engineer in the room.
Match the AI workload to the right platform
One Windows platform, four sizes of on-device AI.
Every seat
Copilot+ PC
Surface Pro and Surface Laptop, plus every OEM Copilot+ PC. Notes, drafting, and search that never leave the device. For the teller, the clinic, and the agency desk that cannot send customer data to a public API.
Installed base · Models to ~13B
The expert
RTX Spark laptop
Surface Laptop Ultra and its OEM peers. One analyst, one machine, one sealed model. Deal files, source code, and case evidence stay in custody, and it works with the network off.
Models to 200B · 128 GB unified memory
The team
RTX Spark desktop
OEM small desktops, plus the Surface RTX Spark Dev Box, which sells through the Microsoft Store only today. A department's private AI under one monitor: contract review, coding assistants, records search.
Models to 200B · 128 GB unified memory
The institution
DGX Station for Windows
From NVIDIA and its OEMs, listed for Q4. The institution's own frontier lab: agent fleets, fine-tuning, and trillion-parameter models inside the building. For regulators, defense, and research where sovereignty is the rule.
Models to 1T · 748 GB memory
Every device in the portfolio runs under one set of rules: Entra, Intune, Purview, Secured-core.
Model sizes are the vendors' published figures as of September 29, 2026, and vary with memory configuration. Product images belong to Microsoft and NVIDIA.
Why us
We help your sellers understand a customer's AI workload, show where local AI fits, and back it up with a working demo.
We spent years bringing Microsoft Surface to market, inside Microsoft and alongside its partners, including Intel and Qualcomm. Today we're independent. We don't sell hardware or software.
We run the hardware and models we teach, every day, and we know their limits. If a workload belongs in the cloud, we'll say so. Meet the founders.
Built around the customer. In Microsoft executive briefings, we built demos customers recognized as their own: their brand, their language, their work. The Accelerator does the same: a working AI proof of concept on the customer's workload, in 30 days. Explore the Accelerator.
April BrickerThe story
Kevin MurphyThe channel
Frank BuchholzThe technology
How it's funded
OEM incentive, co-marketing and proof-of-concept funds can help cover seller training and customer work.
- 1
Scope the work.
We agree on what we'll deliver and how we'll measure success.
- 2
Find the funds.
We work with you and the OEM to find eligible programs.
- 3
Confirm before we start.
Funding depends on availability and approval. We confirm it and spell out any remaining costs before work begins.
- 4
We bill the program.
Cadence 3 is an approved Microsoft supplier, so eligible Microsoft-funded work is billed directly to Microsoft. Less paperwork for you.