Orchestrator Settings
Environment secrets · Provider configuration · API key management
Anthropic Claude — Dual-Provider Consensus & Risk Worker
ANTHROPIC_API_KEY is configured and active. Claude claude-3-5-sonnet-20241022 powers Worker B (Risk & Validation) in the Map-Reduce pipeline and the Level 3 Consensus Verifier (dual-provider inference with OpenAI). Both workers are live — no fallback to OpenAI is occurring.
Provider Status
llama3-8b-8192 (Groq)
claude-3-5-sonnet-20241022
gpt-4o / gpt-4o-mini
gemini-2.5-pro / gemini-2.5-flash
Environment Secrets
3 / 4 configuredAPI keys are stored as server-side environment variables and are never exposed to the browser. To update a key, open your .env file and replace the value, then redeploy. Keys marked Active have real values configured.
Llama 3 8B API Key
Meta / Groq
Required for Worker D (Deep Reasoning) in the Parallel Map-Reduce pipeline and as the neutral Llama 3 8B Verifier Node in Level 3 Consensus. Uses Groq's OpenAI-compatible endpoint (llama3-8b-8192). Without this key, Worker D falls back to Gemini 2.5 Flash and the Verifier Node falls back to OpenAI gpt-4o-mini.
Orchestrator Roles
••••••••••••••••••••••••••••••••••••••••••••••••
Not configured — add to your .env file to activate
Anthropic API Key
Anthropic
Required for the Risk & Validation Worker (Worker B) in the Parallel Map-Reduce pipeline and the dual-provider Consensus Verifier (Level 3). Without this key, both workers fall back to OpenAI.
Orchestrator Roles
••••••••••••••••••••••••••••••••••••••••••••••••
Real value configured in environment — workers are active
OpenAI API Key
OpenAI
Powers the Intent Router (gpt-4o-mini), Fast Track responses, Analytics Worker (gpt-4o), Recommendations Worker (gpt-4o-mini), Synthesizer Node, and Consensus Verifier primary provider.
Orchestrator Roles
••••••••••••••••••••••••••••••••••••••••••••••••
Real value configured in environment — workers are active
Google Gemini API Key
Google Gemini
Used as fallback provider for the Analytics Worker (gemini-2.5-pro) and Recommendations Worker (gemini-2.5-flash) when OpenAI is unavailable.
Orchestrator Roles
••••••••••••••••••••••••••••••••••••••••••••••••
Real value configured in environment — workers are active
Orchestrator Pipeline Wiring
| Worker / Node | Provider | Model | Pipeline Level | Fallback |
|---|---|---|---|---|
| Intent Router | OpenAI | gpt-4o-mini | Level 1 | Heuristic classifier |
| Fast Track | OpenAI | gpt-4o-mini | Level 1 | Gemini 2.5 Flash |
| Worker A — Analytics | OpenAI | gpt-4o | Level 2 Map-Reduce | Gemini 2.5 Pro |
| Worker B — Risk & Validation | Anthropic | claude-3-5-sonnet-20241022 | Level 2 Map-Reduce | OpenAI gpt-4o |
| Worker C — Recommendations | OpenAI | gpt-4o-mini | Level 2 Map-Reduce | Gemini 2.5 Flash |
| Worker D — Deep Reasoning | Llama 3 8B | llama3-8b-8192 | Level 2 Map-Reduce | Gemini 2.5 Flash |
| Synthesizer Node | OpenAI | gpt-4o-mini | Level 2 Map-Reduce | Concatenation |
| Consensus Primary | OpenAI | gpt-4o | Level 3 Consensus | — |
| Consensus Secondary | Anthropic | claude-3-5-sonnet-20241022 | Level 3 Consensus | — |
| Verifier Node | Llama 3 8B | llama3-8b-8192 | Level 3 Consensus | OpenAI gpt-4o-mini |


