How to give production agents room to act without giving them the keys to everything.
The Permission Envelope: A Control Plane for AI Agent Actions
Razvan Mares··8 min read
Production AI agents do not fail because they are too autonomous. They fail because their autonomy is undefined.
An agent that can read a support ticket, inspect an order, issue a refund, update a CR...
PostgreSQLAI/MLrest-apis+2
Advanced AI concepts that turn language models into dependable software components.
The Hidden Runtime of AI Systems: Context, State, and Evaluation
Razvan Mares··8 min read
Large language models are often introduced as an API call: send text, receive text. That mental model is useful for a prototype, but it breaks down quickly in production.
A production AI feature is n...
PostgreSQLci-cdTypeScript+2
A practical way to evaluate .NET 11 across runtime, web APIs, data access, and delivery pipelines.
.NET 11 Preview Guide: Finding the Changes That Matter in Production
Razvan Mares··8 min read
# .NET 11 is not just an upgrade target
A new .NET release can look deceptively simple from a distance: update the SDK, change `TargetFramework` to `net11.0`, resolve a few warnings, and ship. In pra...
ci-cd.NETef-core+2
How to make Argo CD or Flux useful on a small cluster without turning it into a platform project.
Designing a Low-Drama GitOps Control Plane for k3s
Razvan Mares··8 min read
Small k3s clusters reward restraint.
A three-node homelab, a compact edge deployment, or a single virtual machine running k3s does not need to imitate a large enterprise Kubernetes estate. It still b...
ci-cdtraefikgit+4
Why the next phase of Model Context Protocol design is about contracts, trust, and operational discipline.
The 2026 MCP Overhaul: From Tool Demos to Durable Agent Infrastructure
Razvan Mares··9 min read
Model Context Protocol (MCP) began with an elegant promise: give AI applications a standard way to discover and use external capabilities. A model could read a resource, call a tool, follow a prompt t...
nodejsTypeScriptAI/ML+2
A practical guide to running capable language models on your own hardware.
Self-Hosting Local AI Models: Building a Private Inference Stack
Razvan Mares··9 min read
Running an AI model locally changes the relationship between an application and its intelligence. Instead of sending every prompt, document, and conversation to a remote API, the model becomes part of...