Valdr Blog
Thoughts on building with AI agents — from orchestration patterns to team workflows to the changing nature of engineering itself.
Automate the Process You've Learned to Trust
How I turned a trusted agent process into a Valdr workflow, with human checkpoints I can reduce as confidence grows.
September 10, 2026
Your Agent Doesn't Need a Bigger Model. It Needs a Smaller Task.
Every stage in a delivery pipeline retires one ambiguity — and every ambiguity you retire lowers the model tier the next stage needs. How spec, plan, and tasks turn expensive reasoning into cheap, repeatable execution.
July 24, 2026
AI Agents Don't Remove Process. They Punish Bad Process.
AI agents do not make unclear work clear. They make unclear work faster, louder, and harder to trust.
June 25, 2026
Stop Paying the Agent Context Tax
A bug report, two repos, and what changes when an agent can do the first deep dive itself. Valdr 0.2.0 makes the expensive part of engineering — the investigation — reusable.
May 28, 2026
I Became the Integration Layer. So I Built Valdr.
A practitioner's account of what breaks when serious AI workflows scale — and why every serious one converges on the same architecture.
May 23, 2026
Valdr Is the Control Plane. Your Workflow Is the Product.
Valdr does not create leverage by itself. The leverage comes from the packs, agents, capabilities, prompts, review rules, and handoff chains you encode.
May 23, 2026
From Coding to Orchestrating: The Paradigm Shift in Software Engineering
The role of the software engineer is evolving. We're moving from hours at the keyboard to directing AI agents — but only if we solve the consistency problem.
December 20, 2024