Blog | Norsica

Sep 5, 2026 · 8 min

Switching from GPT-5.6 Sol to GPT-6 Astra: Start with Medium Effort

Start with GPT-6 Astra medium: it cost 19% less than Sol high in my Codex comparison. High built better code, but medium found a bug its review missed.

#GPT-6 Astra#GPT-5.6 Sol#Codex CLI

Aug 20, 2026 · 15 min

Was the Monorepo Really Best for Coding Agents?

Repository boundaries stopped being context boundaries for coding agents. The monorepo advantage moved from retrieval to governance, far harder to operate.

#Monorepo#Coding Agent#Context Engineering

Aug 5, 2026 · 17 min

Reasoning Effort Is Not a Quality Setting

Across 28 runs, higher reasoning effort expanded what coding agents explored, not what they delivered. Its value depended on the phase and the model.

#Coding Agent#Reasoning Effort#Agent Workflow

Aug 2, 2026 · 12 min

When Better Models Make Old Agent Workflows Worse

Agent workflow rules built for old model failures start manufacturing work. Keep constraints at boundaries and evidence, and let the model choose the path.

#Agent Workflow#Coding Agent#Boundary Constraint

Feb 11, 2026 · 13 min

What Makes Development AI-Native?

AI-Native development requires four structural pillars most organizations lack. Assess your readiness with a 100-point diagnostic criteria.

#AI-Native#Context Closure#LLM

Jan 26, 2026 · 12 min

Planning Is the Real Superpower of Agentic Coding

Turn LLM execution into verification with explicit planning. Research shows this approach improves task success by 33% in complex agentic coding scenarios.

#Agentic Coding#Planning#LLM

Jan 21, 2026 · 11 min

The Leadership Bottleneck in AI-Native Development

AI productivity metrics look great, but outcomes haven't changed. Discover why the leadership bottleneck, not tools, determines AI-Native transformation success.

#AI-Native#Leadership#Engineering Management

Jan 19, 2026 · 9 min

Stop Putting Everything in AGENTS.md

Learn where to write LLM rules, what to write, and how to verify they work. Keep AGENTS.md lean while ensuring your rules actually stick.

#AGENTS.md#Agentic Coding#Rule Externalization

Jan 12, 2026 · 9 min

Why LLMs Are Bad at 'First Try' and Great at Verification

Stop optimizing for first-shot perfection. LLMs perform better at verifying and improving existing artifacts than at controlled first-time generation.

#LLM#Verification#Workflow Design

Jan 12, 2026 · 9 min

Building a Local RAG for Agentic Coding: From Fixed Chunks to Semantic Search with Keyword Boost

Started with a simple RAG for MCP, then ended up implementing semantic chunking and rethinking hybrid search. For people who've built RAG systems and started hitting quality limits.

#RAG#Semantic Chunking#MCP

Jan 12, 2026 · 17 min

Rethinking Team Development in the Age of LLMs

Code reviews shift from reviewing code to reviewing specs. Design the whole Spec → Plan → Test → Implementation cycle. Small teams work better, so scale by adding teams, not people.

#LLM#Team Development#Process Design