AI in NLP didn’t begin with products.
It began with papers.

Pre-training + transformers showed that one model can learn language once and adapt everywhere.
That idea still powers LLMs, search, and RAG today.

Read the research. Build better systems.

Built this visual to capture a simple idea:
Old databases learned new tricks.
pgvector turns Postgres into an AI-native system.
If you’re building search, RAG, or recommendations —
this belongs in your toolbox.
#BuildInPublic #pgvector #PostgreSQL #AIBuilder #DevLife

2026 = stop chasing AI hype. Start shipping AI that works.

• AI writes a lot of the code now
• Real skill = directing systems, not typing
• Rust is default for secure systems
• Git is your infrastructure
• Agentic AI is hot, but prod-ready is hard

Winners orchestrate.
Losers chase models.