SnowShoe.ai lets technical teams ask questions in plain language and get answers drawn from their own documents, with the source passages attached so a person can verify every answer. It runs inside your infrastructure, including fully air-gapped deployments.
Meaning-based retrieval with re-ranking runs alongside exact keyword search, so conceptual questions and precise term lookups both work.
Each retrieved passage carries the document it came from, so answers are traceable back to your material.
A fully offline installer bundles the models. Query content stays on your network.
PDF, Word, PowerPoint, Excel, CSV, text, Markdown, HTML, and images. Scanned and image-only PDFs are read with OCR.
Domain vocabulary handled explicitly, rather than guessed at.
Scores retrieval accuracy against your own benchmark questions and expert answers, so you can measure before you trust.
Problem, deployment model, licensing, and how this relates to our consulting work, stated plainly.
Your answers already exist, in maintenance manuals, drawings, procedures, and years of accumulated reports. The cost is not that the knowledge is missing, it is that finding it takes an experienced person an afternoon, and that person is the constraint. Generic AI search makes this worse, not better, when it answers confidently from the model's general memory instead of your material.
Retrieval-grounded search over your own document set, where every answer links back to the passage it came from, so a reviewer can check it.
Self-hosted on Kubernetes or Docker Compose. A fully offline installer bundles the models, so it runs with no outbound internet access: on-prem, air-gapped, or inside your own cloud enclave.
Engineering and maintenance operations; defence and aerospace programs; industrial and manufacturing environments; regulated enterprise IT and compliance-sensitive teams.
Most SnowShoe deployments start as a consulting engagement: we scope which document sets matter, run the evaluation harness against your benchmark questions, and integrate the result into how your team actually works. The platform is licensed; that integration work is a normal fixed-fee project.
Perpetual licence per deployment, with annual support and updates. Pilot on a sample of your documents first.
Deployed in engineering and defence environments where the document set cannot leave the network. Air-gapped installation is a supported configuration, not a special case.
Custom engagements. Clients own project-specific code, models, and documentation as defined in the applicable agreement. That is the default for consulting work and it does not change because a project touches one of our products.
Lemay.ai products. Pre-existing Lemay.ai platforms and intellectual property are licensed separately where selected. A licence covers the platform; it does not claim anything you build on top of it.
Tell us what you are trying to accomplish and we will tell you whether this is the right fit, including when it is not.