Sales Lab · Edge
The model lives where your data lives.
Banks, insurance, regulated B2B with PII / PCI / SOC 2 carve-outs. The conversation never leaves your VPC. Edge ships you the model + the runtime + the orchestrator; the control plane is yours.
Inference
Anthropic / Mistral / Llama 3.x on dedicated GPUs you own. We support BYO-GPU + the model bundle, or full-rack drop-in (NVIDIA H200 + 80TB NVMe). Cold-start latency ≤4s; steady-state turn latency p99 under 600ms.
Boundary
The lab never sees a transcript. Your SOC ingests the same telemetry your existing tooling ingests; nothing flows to Sales Lab's SaaS plane. The control plane orchestrates training scenarios + scoring config, not data.
Compliance
SOC 2 Type II + ISO 27001 reports for the orchestrator. Customer-owned KMS for at-rest encryption. CJIS-compatible deploy modes available.
Difference from On-prem
On-prem is fully air-gapped. Edge keeps a thin control-plane connection for orchestration (no data, no transcripts) so you don’t have to manage releases manually.
Pricing
Bring-your-own GPU: $350k/year floor + per-seat at the same Enterprise rate. Full-rack drop-in: starting $750k year one (hardware + install) + $200k year two onward.