00PROCAD EXPO 2026 · GDAŃSK · 15-16 OCT

Your AI, inside your boundary

AI agents on local models work in browser-based Windows workspaces - inside Autodesk tools, through an MCP server. Your data never leaves your infrastructure, and the cost is fixed.

32 GB VRAM GPUAutodesk MCPNo tokens
hex_bolt_M12x40.dwg[Model]
X 0.00 Y 0.00 Z 0.00
Command:
01Problem

Your data must not leave the company

More and more projects - from infrastructure and energy to industrial and public buildings - require engineers, architects and designers to guarantee that data never leaves their infrastructure. Public cloud AI tools cannot give that guarantee.

CONFIDENTIALITY[01]

The client demands a guarantee

Contracts, tenders and NDAs increasingly forbid moving project data outside the contractor's infrastructure.

DATA[02]

Cloud AI sends it outside

Every prompt with a drawing, model or specification ends up on an outside provider's servers.

CONTROL[03]

You don't know what happens to it

Where it is processed, who can access it and how long it is kept - all outside your control.

COST[04]

The bill grows with use

You pay for every token, and an agent working all night inflates the invoice.

02At the booth

Watch an agent at work at Procad Expo

At our booth we run AI agents in our workspaces. Everything runs locally: from the browser, through Windows, to the model on the GPU.

32 GBVRAM on the AI server
Windowsin the browser
0tokens sent to the cloud
  1. Open a workspace in the browser

    A full Windows desktop with nothing to install - Autodesk tools preinstalled, including Fusion.

  2. Give the agent a task

    Describe it in plain language, for example: design an M12×40 bolt.

  3. The agent works in Autodesk tools

    It connects to the official Autodesk MCP server and carries out each step.

  4. The model runs on a local GPU

    An AI server with 32 GB of VRAM. No data goes to the cloud.

03Live

Watch the AI server at work

You type the prompt in our workspace at the booth, and here you watch the AI server live: GPU load, memory and generation speed. Every task the agent runs shows up on the chart.

grafana · gpu-01live
GPU6%
VRAM11.2 GB
generation0 tok/s
04Platform

AI that works for you 24/7

01

Browser-based workspaces

Windows and Linux in the browser - for people and for AI agents working around the clock.

02

Local GPU/AI server

Compute for your whole team, on your own infrastructure.

03

AI agent marketplace

Ready-made agents for different jobs, or your own, tailored to your needs.

04

MCP integration

Connects to tools through the Model Context Protocol, including Autodesk MCP servers.

05

Your data stays with you

Projects, prompts and results never leave your network.

06

Fixed cost

You pay for the server and workspaces, not for tokens. The bill does not grow with use.

05Workspaces

Virtual workspaces, on-prem too

Our virtual machines can run on your own infrastructure, next to the AI server. Every employee gets their own isolated workspace where an agent works around the clock.

[01]

On-prem or in the cloud

Workspaces can run on your own servers, on the same network as the local AI model.

[02]

An isolated workspace for everyone

Every employee and every agent gets their own separate machine.

[03]

Agents work 24/7

Tasks keep running even when the employee shuts down their work computer.

[04]

Access from the browser

A full desktop in the browser, with nothing to install on the laptop.

[05]

Multiplayer and communication

Work together in one workspace, with built-in communication features.

[06]

Windows or Linux

Pick the system for the tools and the job - Autodesk, Fusion or scripts.

An extra layer of security and isolation

The agent doesn't run on the employee's computer but in a separate workspace. Project data stays on the server, not on laptops.

06Who it is for

For teams that design and build

The agent takes over repetitive work in the model, while your team stays focused on design decisions.

[01]Architects

The agent prepares variants, repeated elements and notes - you focus on the concept.

[02]Structural and mechanical engineers

Parametric parts, connections and schedules generated from a plain-language description.

[03]MEP designers

Routine notes, material schedules and revisions without manual clicking.

[04]BIM managers

One controlled environment: models, access and logs inside the company boundary.

[05]Drafters

Dimensions, notes and title blocks filled in by the agent.

[06]IT teams

The AI server and workspaces on your infrastructure, with control over access and network traffic.

07Why on-prem

Security. Fixed cost. Compliance.

1

Security

Models, workspaces and data run inside your organization's boundary. Outbound traffic is blocked by default.

2

Fixed cost

An agent working all night does not inflate the bill. You know the price upfront.

3

EU AI Act compliance

AI Act obligations phase in until 2028. Control over data, access and logs makes them easier to meet.

[0,0][620,0][1240,0]
08Contact

Let's talk

Book a meeting

30 minutes: we'll talk about your scenario and discuss running it on your infrastructure.

Prefer to call? +48 787 255 031

Open in cal.com

The cal.com calendar sets its own cookies, so we only load it with your consent.

Waitlist

Sign up and we'll let you know when Teamsharq AI is available for your team.