Nace introduces Console, a platform for executing AI-agent workflows through one API key and a shared HTTPS and JSON interface. It combines Drex, a small decision model that assigns probabilities to options in a single forward pass, with Nace Document Intelligence (NDI), which processes 36 file types and points extracted values and answers back to source pages or locations. Drex is positioned for routing, ranking, classification, guardrails, tool selection, retries, and next-action decisions, while NDI supports document splitting, classification, schema-based extraction, OCR, transcription, named-entity recognition, and grounding. The page shows REST endpoints for decisions and document jobs, bearer-key authentication, idempotency keys, SDKs for TypeScript and Python, CLI and MCP access, and integrations tested with coding and agent tools including Codex, Cursor, Claude Code, Hermes, and opencode. Nace presents four deployment options: managed cloud, dedicated cloud, deployment inside a customer’s AWS or Azure account, and on-premises or air-gapped installation. It says the same models, API, and serving stack can be used across these boundaries, while customers retain their weights, adapters, and evaluation sets within the selected environment. The page also states that the managed cloud does not train on customer data and describes customer-managed keys, logs, IAM, private links, or offline updates for more controlled deployments.
