Updated VectorDB ADR to reflect ChromaDB as a provisional decision

docs/adr/0001-vectordb-backend.md

@@ -1,43 +1,50 @@
-# 1. VectorDB
-
-Contents:
-
-- [Summary](#summary)
-  - [Issue](#issue)
-  - [Decision](#decision)
-  - [Rationale](#rationale)
-  - [Status](#status)
-- [Context](#context)
-  - [Important Factors](#important-factors)
-  - [Options](#options)
-  - [Supporting Information](#supporting-information)
-  - [Assumptions](#assumptions)
-  - [Constraints](#constraints)
-  - [Implications](#implications)
+# VectorDB
+
+## Table of Contents
+
+- [VectorDB](#vectordb)
+  - [Table of Contents](#table-of-contents)
+  - [Summary](#summary)
+    - [Issue](#issue)
+    - [Decision](#decision)
+    - [Rationale](#rationale)
+    - [Status](#status)
+  - [Context](#context)
+    - [Important Factors](#important-factors)
+    - [Options](#options)
+    - [Supporting information](#supporting-information)
+      - [Benchmarks](#benchmarks)
+    - [Indexing](#indexing)
+    - [Assumptions](#assumptions)
+    - [Constraints](#constraints)
+    - [Implications](#implications)
 
 ## Summary
 
 ### Issue
 
-We need to support a vector db for retrieval-augmented generation (RAG), there are many options but we need to choose between them. 
+We need to support a Vector DB for retrieval-augmented generation (RAG), there are many options but we need to choose between them.
 
-### Decision 
+### Decision
 
-TBD
+ChromaDB
 
-### Rationale 
-TBD
+### Rationale
+
+We decided on ChromaDB due to it's simple implementation, performance, and Apache 2.0 license. At this maturity level, we expect individual capabilities on top of LeapfrogAI to maintain their own Vector DB with ChromaDB and will expand support as common themes surface.
 
 ### Status
+
 Provisional
 
 ## Context
 
 ### Important Factors
+
 - OpenSource
 - Performance
-    - Queries per second
-    - Latency
+  - Queries per second
+  - Latency
 - RBAC
 - Self-hosted
 - Portability
@@ -49,7 +56,9 @@ Provisional
 
 - [Pinecone](https://www.pinecone.io/)
 - [Weaviate](https://weaviate.io/)
-- [Milvus](https://milvus.io/) - [GitHub](https://github.com/milvus-io/milvus) - [Architecture](https://milvus.io/docs/architecture_overview.md)
+- [Milvus](https://milvus.io/)
+- [GitHub](https://github.com/milvus-io/milvus)
+- [Architecture](https://milvus.io/docs/architecture_overview.md)
 - [Qdrant](https://qdrant.tech/)
 - [Chroma](https://www.trychroma.com/)
 - [Elasticsearch](https://www.elastic.co/elasticsearch/)
@@ -64,13 +73,13 @@ Provisional
 - https://github.com/milvus-io/milvus/discussions/4939
 
 ### Indexing
+
 - https://weaviate.io/blog/ann-algorithms-vamana-vs-hnsw
 - https://thedataquarry.com/posts/vector-db-3/
 
-
 ### Assumptions
 
-- Choosing a single vector db is desirable
+- We will start with a single preferred VectorDB, but LeapfrogAI should be extensible to accommodate other options.
 - Not bundling the vector db with the app layer is desirable
 
 ### Constraints