V
Victorious Chimpanzee
Currently, Harness AI features for Continuous Delivery (CD) on Self-Managed Enterprise Edition (SMP) rely on a fixed external endpoint (https://harness.openai.azure.com). For organizations running SMP, particularly in regulated industries or air-gapped/offline environments, this creates a blocker, as security and compliance policies often require the use of internally approved LLMs rather than external SaaS-hosted models.
Harness Agents already support a BYOM model via LLM Connectors, allowing teams to connect Anthropic, OpenAI, or Gemini, including self-hosted model endpoints. We would like to see this same capability extended to Harness AI features in CD on SMP, so that teams can:
  • Connect a company-approved or self-hosted LLM for AI-powered root cause analysis and deployment troubleshooting in CD pipelines
  • Use a custom endpoint (e.g., an internal model deployment) via the existing AIModel connector type
  • Maintain full compliance with internal data residency and security policies
This would bring SMP customers to feature parity with SaaS for AI-driven CD capabilities, while respecting enterprise governance requirements.