pgvector. The retrieval stack you already host.
Moonlabs is the operator-led AI Academy in Derby. We run three live companies — Homemove, home.co.uk and homedata.co.uk — and we teach twelve students per cohort to ship a real AI product, sell it to a real customer, and raise on it. Three pillars: Coding, Commercials, Investment. Twelve weeks. £6,000.
Moonlabs is what we are. Two operators — James Freestone and Louis O’Connell-Bristow — who run Homemove, home.co.uk and homedata.co.uk. Postgres + pgvector is our default retrieval stack across all three companies — the indexing decisions and SQL patterns on this page are the ones we maintain in production, not slides from a vendor pitch.
The Academy is what we do. A twelve-week, in-person, twelve-student cohort in Derby. You build a real AI product. You sign a paid pilot on it. You write a deck and a financial model. You leave with a deployed system, a paying customer reference and a live investor pipeline. Coding, Commercials, Investment — the three pillars taught in equal weight every week.
Why this page exists. Most production RAG systems do not need a dedicated vector database. Pinecone, Weaviate and the rest are real products, but the honest truth is that for an enormous proportion of real-world retrieval workloads, pgvector inside the Postgres database you already run is good enough — often faster, almost always cheaper, infinitely easier to operate. Useful supporting essay here. You leave the Academy with a pgvector-backed retrieval system in production — or as the founder of a vertical-search product that ships on the database the customer already has.
Coding · pgvector at production fluency
HNSW vs IVFFlat index strategy, distance functions, dimensionality, hybrid lexical + vector in a single SQL plan, reranking, maintenance windows on rebuilds, trade-offs at 1M / 10M / 100M vectors. Retrieval-level evals (recall@k, MRR, precision-by-vertical) from week one. A deployed pgvector-backed retrieval system by week twelve.
Commercials · pgvector as a productised offer
Almost every SMB on Postgres has a knowledge base they could turn into a transformative AI surface — without paying a vector-DB SaaS bill. Pricing a pgvector retainer, the discovery call, a one-page pilot agreement, the first paying customer. A paid pilot by week six — for many graduates, the “cheaper retrieval than your incumbent SaaS” pitch lands fast.
Investment · raising on Postgres-AI infrastructure
Supabase ($2bn, Postgres + pgvector first), Neon ($3bn now Databricks), Tembo, Crunchy Data acquired by Snowflake at $250m, Xata, Nile — Postgres-AI is one of the loudest infrastructure thesis areas in venture. Cap table, ten-slide deck, financial model. A live investor pipeline by demo day.
Common questions.
Is pgvector really enough for production?
For the majority of production retrieval workloads in 2026 — yes. The crossover point where a dedicated vector database earns its place is materially higher than most vendors will admit. The essay covers the numbers honestly.
When should I use Pinecone / Weaviate / Qdrant instead?
When you have hundreds of millions of vectors and serious low-latency requirements; when you need geographically-distributed read replicas the dedicated vector databases handle better; when the operational simplicity of a managed vector store beats the operational simplicity of one less database to run. Most teams do not actually have those constraints.
How does this fit with the RAG curriculum?
pgvector is the default retrieval stack inside the wider RAG curriculum. This page is the specifically-Postgres entry point.
Will I learn LlamaIndex or LangChain on top of pgvector?
Yes, where they earn their place. LlamaIndex and LangChain both run cleanly on top of pgvector; we teach the integration patterns that survive a production deploy.
Do I need Postgres experience?
Helpful but not required. We will get you fluent in the Postgres bits that pgvector cares about inside the first fortnight; the wider Postgres curriculum is folded into the project work.
More Academy entry points.
The Academy is one course with many doors. Each of these pages is a different entry point into the same twelve weeks.
Build it. Sell it. Raise on it. In twelve weeks.
Tell us what you would retrieve over and what shape the Postgres database is in already. James and Louis read every application personally and reply inside the week.
© 2026 Moonlabs Incubator. All rights reserved.