Measure whether the grounded café FAQ application works

A plausible result is not yet evidence. Evaluate with golden-question pass rate, citation support, abstention accuracy, response time, and a clean-room demo by another person. The test fixture should contain an easy positive case, an easy negative or baseline case, and the boundary case most likely to flip. Separate assertions about software contracts from claims about model quality: both matter, but they answer different questions.

The artifact's user-facing goal is specific: combine input validation, five-note retrieval, a cited answer, tests, logs, and a two-minute demo into one finishable application. Its accepted input is a user question plus a reviewed local FAQ collection with stable source IDs. Those statements are intentionally narrower than “build an AI system.” Narrow scope lets us inspect every input and expected result, and it prevents a toy result from being presented as a production claim. The runnable check below turns one success criterion into an assertion, so a regression exits loudly.

Save this as lesson.py and run python3 lesson.py. It uses only the language standard library, so the example is reproducible offline.

gold=[('When is the cafe open?','hours'),('How do I get wifi?','wifi')]
docs={'hours':'cafe open weekdays','wifi':'get wifi card'}
def hit(q): return max(docs,key=lambda k:len(set(q.lower().split())&set(docs[k].split())))
assert all(hit(q)==want for q,want in gold); print('hit_rate',1.0)

Expected output: hit_rate 1.0. Exact floating-point formatting may vary slightly, but the asserted behavior must not. Read the output as evidence about this stage, not merely proof that the interpreter started.

Walk one question through validation, tokenization, score table, selected FAQ ID, cited answer, and safe event record. If the clean-room demo fails, resist adding setup prose until you reproduce the missing assumption. Unsupported questions must reach the abstention branch. Treat a source containing instructions as data and sanitize or reject it during review.

At the evaluation stage, save the smallest failing fixture beside the expected result. Change one cause at a time and rerun the exact command printed above; that makes the repair reviewable and keeps this chapter's progressive artifact reproducible.

Run golden questions for every FAQ note and unsupported questions for abstention, then review citation support by hand. Time the two-minute demo from a clean checkout and ask another person to follow it without coaching. Record the FAQ version, passing counts, response time, owner, and tested rollback snapshot. Those are the capstone's shipping claims.

For this evaluation page, preserve the fixture and result as evidence for the next page. Label observations separately from conclusions: a passing assertion establishes the behavior it names, while broader usefulness requires the chapter's full evaluation set and stated operating limits.