Why One Model Should Not Handle Every AI Task
Using one model for every AI task turns model choice into a hidden default. Production workflows need explicit model policies for task type, risk, latency, evidence, fallback, and evaluation.
Read more
Blog
Foundations and patterns for working AI systems.
The general blog holds the reusable ideas: foundation beliefs, system-layer patterns, decision frameworks, production lessons, and project lessons that travel beyond one build.
Essay library
Build beyond the demo.
These posts are the connective tissue around the projects: how to think, what to test, what usually breaks, and which system layers should exist before production pressure arrives.
Token Economics: Why Prompt Bloat Kills AI Margins
Token waste is architecture debt. Oversized prompts, broad context, verbose outputs, retries, review, and eval runs compound into the cost per trusted completed unit.
Read more
Semantic Caching Is Harder Than It Looks
Semantic similarity measures intent, not validity. Safe AI caching needs source versions, schemas, model policy, evidence, permissions, review state, and wrong-hit tracking.
Read more
How to Trace a Failed LLM Run
When an AI workflow fails, the final answer is not enough. You need the input, route, prompt, model, context, schema errors, evidence gaps, review action, and eval case.
Read more
Designing Fallbacks for AI Systems
AI failures need designed paths: retry, narrower scope, simpler model, stronger model, human review, partial answer, safe message, or stop.
Read more
What Happens the Day After an AI System Launches?
Launch is not the finish line for an AI system. It is when review queues, failures, costs, user feedback, model changes, and ownership finally become real.
Read more
Code-First AI Systems: Why the Model Should Not Run the Workflow
The model should classify, extract, summarize, judge, or propose. Code should own routing, state, validation, retries, evidence, review, and evaluation.
Read more
Pydantic as an AI Architecture Boundary
LLM output is not application state until it passes a typed boundary. Schemas, validation errors, evidence fields, and versioned contracts make AI workflows easier to inspect and change.
Read more
The Death of the 1,000-Line Prompt
Giant prompts feel productive at demo stage. In production, workflow logic needs to move into code, schemas, routing, validation, evidence, review, and evaluation.
Read more
Why Chunking Breaks Business Documents
Chunking is not just a retrieval setting. In business documents, the wrong chunks can separate values from labels, tables from headers, clauses from definitions, and evidence from the answer.
Read more
Why PDF Extraction Is Harder Than It Looks
PDF extraction fails in production because business documents are not clean text. They carry layout, tables, repeated fields, scans, conflicts, and evidence requirements.
Read more
Evaluation Is an Engineering Problem
Why AI evaluation is not a report card after launch, but a design constraint from day one.
Read more