aisle

Learn. Plan. Design.

Plan on-prem AI inference infrastructure with confidence.

Aisle - AI Sizing and Learning Environment - is the planning workbench for IT teams. Walk every layer of the inference stack and build a sizing you can defend.

Open source · Vendor neutral · Runs in your browser

What happens when AI generates one token?

Interactive

User

prompt

API Gateway

auth, rate limit

Inference Server

batch, schedule

GPU · HBM

Weights70 GB
KV CacheDynamic
Headroom~25%
PromptWhat is the capital of France?
Output(empty)|

Pick where you want to start.

Four ways into the aisle.

  1. New to AI infrastructure?

    Start with the fundamentals.

    Short educational modules to learn what inference actually is, what components are involved, and how to size and plan an on-prem deployment.

    Start learning

  2. Have a workload to size?

    Open the Sizer.

    Tell us your model, traffic, and latency SLOs. Get a server-spec recommendation across baseline, burst, and resilient scenarios. Shareable URL, exportable summary.

    Plan a deployment

  3. Exploring the inference stack?

    Browse the layers.

    Six layers of an on-prem inference deployment, from the GPU to the rack. Two worked examples (enterprise scale and departmental scale) show how the pieces fit together.

    Explore components
  4. Coming soon

    Already have a sizing?

    Visualize the topology.

    Turn a sizing into a deployment archetype with a topology diagram. Picks between single-node, multi-GPU, and multi-node patterns based on your inputs.

    Open Designer