• Ollama standard and Open WebUI default
• RTX 4000 Ada class 20 GB model-fit shortlist
• Driver, storage and service readiness check
• Benchmark report before throughput promises
• Managed updates, monitoring and backups
• Email and ticket support

Private managed AI for teams that want local model hosting without a third-party AI API by default. Live inference claims stay gated until the actual server passes GPU, Ollama, and target-model smoke tests.

• Ollama and Open WebUI
• Knowledge/RAG setup for team documents
• RAG Retrieval Quality Audit fast path with fixed questions and cited-answer review
• Qwen3-Embedding 0.6B/4B and Qwen3-Reranker 0.6B/4B first, 8B only after fit checks
• Visual RAG and evidence search for screenshots, scans, diagrams and product images
• Qwen3-VL-Embedding and Qwen3-VL-Reranker benchmark option
• User roles, source citations and private domain setup
• Benchmark-first rollout plan with prioritized support

For teams that need source-backed internal knowledge and visual evidence workflows with managed operations, role boundaries, source citations, and a measured retrieval audit before monthly production scope. Live inference claims stay gated until the actual server passes GPU, Ollama, and target-model smoke tests.

• Private document workflow trial
• Docling/OCR and vision-language benchmark
• Security hardening and OIDC/SSO preparation
• Audit logging and retention planning
• Planned update windows
• Monthly review and change-control support

For production business use where private forms, PDFs, invoices, code, or knowledge workflows need security, evidence, review gates, and operating discipline before automation promises.

• Management only for customer-owned or customer-rented GPU servers
• Driver and Ollama readiness check
• Open WebUI installation
• Document/RAG/code workflow benchmark option
• Updates, monitoring and support
• Hosting costs paid by the customer directly

For customers who already have suitable GPU infrastructure and want managed local AI operations without replacing their server or cloud provider.