Private deployment architecture
Deploy inside your AKS/Kubernetes cluster or on-prem servers with explicit network egress and AI provider boundaries.
Enterprise AI Engineering
Built for procurement, security & production reality.
For CTOs, CISOs, VPs of Engineering, and SRE leaders evaluating autonomous AI in production. We provide complete clarity on where our software runs, how authority is enforced, and how deployment fits your operational requirements.
Deployment considerations
Plan your deployment around the workflow.
To help technical evaluators and procurement teams conduct accurate due diligence, this table unifies our commercial platform capabilities (devheallabs.com) and open-source compiler stack (NC, NC UI and NC AI) in one place.
| Capability Domain | Product / Layer | Implementation Scope & Verification Detail |
|---|---|---|
| Customer-Hosted Kubernetes Deployment | SteadyState | Deploys on customer-managed Kubernetes via our AKS deployment kit or Docker Compose for local evaluation. |
| Enterprise Identity, RBAC & 2-Person Approvals | SteadyState + Security | Microsoft Entra ID sign-in, group-based role mapping, separation of duties, and two-person approval gates support controlled access to SteadyState workflows. |
| Hash-Chained Audit Trails & Traces | SteadyState & D-AgentProving | Tamper-evident SHA-256 hash-chained audit logs in SteadyState and signed Merkle-root Certificates of Assurance in D-AgentProving. |
| Single-Binary Self-Hosting & Offline CPU AI | NC · NC UI · NC AI | Dependency-free native binary compilation and local CPU inference with zero cloud account or GPU requirement. |
| Adversarial Agent Proving Ground | D-AgentProving | Digital-twin chaos injection, tournament comparison arena, and scorecard receipts for custom or framework agents. |
| Multi-Agent Observability & Self-Healing | HiveANT · SwarmOps | Operational signal correlation and swarm diagnostics for NC and SteadyState workflows. |
| Enterprise Digital Twin Simulation · Coming soon | D-TwinVoke | Counterfactual blast-radius simulation and invariant contract verification prior to production invocation. |
Enterprise engagement model
A practical path from pilot to production.
Work with our engineering team to scope deployment boundaries, identity integration, and measurable success criteria.
Workflow & toolchain integration
Connect PagerDuty, Prometheus, Kubernetes APIs, Slack/Teams, and internal runbooks without exposing production credentials to public clouds.
Governance & authority design
Configure dry-run defaults, change-freeze calendars, namespace boundaries, and two-person Entra ID approval policies.
Pre-production agent proving
Run candidate agents through D-AgentProving stress suites to verify tool accuracy, recovery, and cost bounds before rollout.
Baseline & outcome measurement
Establish pre-deployment baselines for MTTR, recommendation acceptance, manual responder hours, and audit completeness.
Operational handover & SLA
Document cluster ownership, escalation contacts, support hours, and runbook maintenance in a clear engagement agreement.
Work with DevHeal
Start with a scoped enterprise architecture review.
Bring your security, SRE, and AI architecture leads. We will map one priority workflow to your infrastructure and governance requirements.