Writing
Verified, Not Trusted: Expense Analysis, Part Four
Evals measure reliability with reference keys, code graders, model judges and pass-rate gates. Expense examples show what to check and how structured output expands coverage.
Topic
Applied AI Engineering
Designing, implementing and evaluating applications that use AI models. Workflow control, bounded model judgments and verification determine how model contributions become reliable application behavior.
7 pages
Writing
Verified, Not Trusted: Expense Analysis, Part Three
Evals are acceptance tests for prompts and skills: repeated isolated runs, a pass rate for each check, and a minimum pass rate as the gate. Code grades extracted facts and traces failures.
Writing
Verified, Not Trusted: Expense Analysis, Part Two
Imperative Python controls expense analysis and verifies model contributions through source evidence, reconciliation, and exact calculations. Models handle bounded judgments.
Writing
Verified, Not Trusted: Expense Analysis, Part One
Expense totals need verifiable source coverage and reproducible calculations. Two prompts show how parsing, category rules, refunds, and code shape the evidence.
Writing
Jev: A Practical Reference
Jev makes bounded semantic judgments with Choice, Score, and Noul; code handles exact work and policy, while evaluation and confidence gates determine when to act or escalate.
Writing
Using Jev Compact LLM Context
Jev can shorten LLM context by scoring which tool events a current question still needs. A recorded Noul response and Python retention code show what stays and why the threshold matters.
Writing
Skills versus Controlled Workflows
Stakes and human verification determine where workflow control belongs. Explicit prompts guide a model; imperative code enforces steps, checks, and use of results.