Feature 1 — Integrate LangChain4j & Maintain OpenAI Backward Compatibility

[fmontes]

Outcome:

• Add LangChain4j as a dependency.
• Create LangChain4jClientFactory with OpenAI only.
• Refactor AIProxyClient to use the new factory behind a toggle.
• Validate completions, images, embeddings — same results as before.
• Backward compatibility: no config changes required for current OpenAI customers.

Additional Spike Task in Sprint 1:

• Spike: Research & prototype how dotAI config will handle multiple provider-specific secrets and settings.

• Dynamic config vs. multiple apps.

• Where secrets live (per host, environment).

• Short doc + prototype: switch ProviderType → correct config loads.

Why:

• Proves the LangChain4j abstraction works.
• No regressions for existing OpenAI customers.
• Gets early answers to the biggest architecture risk — without blocking delivery.

---

Related Epic: #32634

[github-actions]

PRs linked to this issue

• #32635 adding multivendor configuration first step by @jdotcms

PRs linked to this issue

• #32635 adding multivendor configuration first step by @jdotcms
• Issue 32635 ai multivendor parser by @jdotcms

PRs linked to this issue

• #32635 adding multivendor configuration first step by @jdotcms
• Issue 32635 ai multivendor langchain it by @jdotcms
• Issue 32635 ai multivendor parser by @jdotcms