Featured AI engineering
Hybrid retrieval: why pure vector search fails in production
Dense embeddings miss the exact terms business users type. What we change when a RAG prototype has to become a system people rely on.
Web7 insights
Engineering notes.
What we learn building AI systems, ERP platforms, applications and the infrastructure underneath them. No thought leadership, just what worked.
Latest articles
-
ERP
What to get right before an ERP go-live
ERP projects rarely fail because the software could not do the job. They fail at go-live. The checklist we insist on before switching a business over.
-
Security
A security baseline for growing businesses
Most breached businesses were not targeted by anyone sophisticated. The baseline that removes the majority of that risk, in the order we do it.
-
Startups
Shipping an MVP without shipping technical debt
Which parts of engineering quality are cheap now and expensive later, and which ones you can genuinely defer before product-market fit.
-
AI engineering
Observability for AI systems: what to log and why
AI features fail softly, plausibly and without an exception in the logs. The records and alerts that tell you before your users do.
-
Cloud & infrastructure
Cloud, on-premise or hybrid: choosing where your systems run
The hosting model matters less than the operational questions underneath it. The requirements that decide it, and the setup we insist on either way.
