{"asOf":"2026-10-28","storesIpAddresses":false,"archetype":"rag-production-stack","requestedArchetype":"rag-production-stack","budgetAppliedUsd":null,"title":"High-Precision Production RAG & Document Intelligence Stack","summary":"Layout-aware PDF parsing, high-recall embeddings, hybrid vector + BM25 search, reranking, and faithfulness grading.","estimatedMonthlyUsd":{"hobby":0,"startup":95,"scale":780},"oneLineInstallCmd":"npm install reductoai voyageai @qdrant/js-client-rest cohere-ai braintrust","requiredEnvVars":["REDUCTO_API_KEY","VOYAGE_API_KEY","QDRANT_URL","QDRANT_API_KEY","COHERE_API_KEY","BRAINTRUST_API_KEY"],"wiringNotes":"Ingest PDFs via Reducto -> embed with Voyage -> upsert into Qdrant Cloud -> rerank top-25 to top-5 with Cohere Rerank.","slots":[{"role":"Document & Table OCR Parser","jtbd":"document-parsing","providerId":"reducto","providerName":"Reducto","editorialOverall":8.02,"activeSku":"reducto-parse","envVars":["REDUCTO_API_KEY"],"smokeTestCmd":"test -n \"$REDUCTO_API_KEY\" && echo \"REDUCTO_API_KEY configured\"","day2Gotcha":"Extract/Edit SKUs get expensive vs LlamaParse Fast/Cost-effective","whyItFits":"Preserves complex multi-column tables and bounding boxes for citation grounding.","decideUrl":"/api/decide?jtbd=document-parsing"},{"role":"Code & Text Embeddings","jtbd":"embeddings","providerId":"voyage","providerName":"Voyage AI","editorialOverall":8.4,"activeSku":"voyage-3-large","envVars":["VOYAGE_API_KEY"],"smokeTestCmd":"curl -sS https://api.voyageai.com/v1/embeddings -H \"Authorization: Bearer $VOYAGE_API_KEY\" -H \"Content-Type: application/json\" -d '{\"input\":[\"ping\"],\"model\":\"voyage-3-lite\"}'","day2Gotcha":"Narrower ecosystem than OpenAI/Cohere embeddings","whyItFits":"Top retrieval NDCG across technical, legal, and code corpora.","decideUrl":"/api/decide?jtbd=embeddings"},{"role":"Managed Hybrid Vector DB","jtbd":"managed-vector","providerId":"qdrant","providerName":"Qdrant","editorialOverall":8.59,"activeSku":"qdrant-cloud-hybrid","envVars":["QDRANT_URL","QDRANT_API_KEY"],"smokeTestCmd":"curl -sS \"$QDRANT_URL/collections\" -H \"api-key: $QDRANT_API_KEY\"","day2Gotcha":"Create payload indexes (createPayloadIndex) on tenant_id and jtbd before running filtered vector searches.","whyItFits":"Rust engine with native sparse+dense hybrid search and payload filtering.","decideUrl":"/api/decide?jtbd=managed-vector"},{"role":"Cross-Encoder Reranker","jtbd":"reranking","providerId":"cohere","providerName":"Cohere","editorialOverall":7.95,"activeSku":"rerank-v3.5","envVars":["COHERE_API_KEY"],"smokeTestCmd":"curl -sS https://api.cohere.com/v2/models -H \"Authorization: Bearer $COHERE_API_KEY\"","day2Gotcha":"Not a frontier general chat leader","whyItFits":"Rerank-3.5 boosts top-5 precision before stuffing the LLM context window.","decideUrl":"/api/decide?jtbd=reranking"},{"role":"RAG Faithfulness & Hallucination Eval","jtbd":"rag-evals","providerId":"braintrust","providerName":"Braintrust","editorialOverall":8.23,"activeSku":"braintrust-pro","envVars":["BRAINTRUST_API_KEY"],"smokeTestCmd":"curl -sS https://api.braintrust.dev/v1/project -H \"Authorization: Bearer $BRAINTRUST_API_KEY\"","day2Gotcha":"Broader tracing competitors exist","whyItFits":"Scores context recall and answer faithfulness on every PR.","decideUrl":"/api/decide?jtbd=rag-evals"}],"cite":"Agent Provider Depot Blueprint (rag-production-stack, as of 2026-10-28). https://getagentdepot.com/blueprints"}