Looking for a VectorDB solution on AWS?

Do you want a managed service solution for your Vector DB on AWS? Then check out AWS OpenSearch.

Just a refresher VectorDB’s or at least Vector indexes are great tools for indexing unstructured data or just massive blocks of text.

For example I am playing around with making all Schematical’s internal SOPs and all my public writings available for my LLM Agents to be able to read and pull from.

Perhaps someday I will release an agent… though considering my crusade against slapping a chatbot on every website and calling it “AI” maybe I shouldn’t.

OpenSearch offers both a provisioned hourly version and a serverless version. If you want to play with some provisioned instances they do have a Micro instance you can play around with for about $13/month.

What really excited me about this is that they seem to have a really flushed out Sharding/Partitioning Strategy so you can scale horizontally.

What is really interesting is that ElastiSearch released a Vector DB that I figured AWS ElastiSearch would have adapted but I think the makers of ElastiSearch must have changed their licensing before releasing the Vector indexing functionality so its not on AWS quite yet.

Either way I am really excited to point my Agents, which are currently using MongoDB as a VectorDB, at some sharded AWS OpenSearch instances to see what they can do.

Question for you:

Have you played around with OpenSearch at all yet?

PS: If you are looking to jam on tools like Vector DBs you should join our Thursday morning AI Jam at 10AM Central US time.