Expected results
These are the committed sample runs, graded by pharmacy_demo. Your own runs in Labs 05 and 10 will differ in wording and sometimes by a point per question. What should hold: the AI-ready model scores clearly higher, and Q02, Q04, and Q05 improve the most.
Scores
Section titled “Scores”| Question | Before (baseline) | After (AI-ready) |
|---|---|---|
| Q01: How many prescriptions were filled by region and month? | 2 (matches the expected answer) | 2 (matches the expected answer) |
| Q02: Which medication categories experienced the largest volume changes? | 0 (does not match) | 2 (matches the expected answer) |
| Q03: What was the average prescription processing time by store? | 1 (partial (top_key)) | 2 (matches the expected answer) |
| Q04: Which stores had unusually high refill activity? | 0 (does not match) | 2 (matches the expected answer) |
| Q05: How did prescription volume compare with the previous period? | 0 (does not match) | 2 (matches the expected answer) |
| Total | 3 of 10 | 10 of 10 |
Baseline run notes
Section titled “Baseline run notes”Illustrative: typical mistakes against the baseline model (missing relationship, summed minutes, wrong period, count instead of rate). Not live agent output.
| Question | Assumptions the agent recorded |
|---|---|
| Q01 | rgn is the region column.; yr_mth is year and month. |
| Q02 | ther_cat is the medication category.; Latest month is 2025-12. The identical values per category weren’t investigated. |
| Q03 | proc_mins summarizes by sum in the model, so the default aggregation was used. |
| Q04 | fill_typ R means refill.; High activity means the highest refill count, all dates. |
| Q05 | Previous period means the previous year. |
AI-ready run notes
Section titled “AI-ready run notes”Illustrative: answers from the AI-ready model using its measures. Not live agent output.
| Question | Assumptions the agent recorded |
|---|---|
| Q01 | Count every fill, new prescriptions and refills together, by store region and calendar month for all 24 months in the data (January 2024 to December 2025). |
| Q02 | Compare fills in the latest complete month in the data (December 2025) with the previous month (November 2025) for each therapeutic category. Rank categories by the absolute change in fills, largest first. Report both months, the change, and the percent change as a fraction (0.25 means 25 percent). |
| Q03 | Average processing minutes per fill for each store across the full data window (January 2024 to December 2025). Sort slowest first. |
| Q04 | For the latest quarter in the data (October to December 2025), compute each store’s refill rate (refills divided by all fills) and its index against the chain-wide refill rate for the same quarter. A store is unusually high when its index is 1.25 or more. Return only the flagged stores. Rates are fractions (0.8 means 80 percent). |
| Q05 | Compare total fills in the latest complete month in the data (December 2025) with the previous month (November 2025). Report the period as YYYY-MM, both totals, the change, and the percent change as a fraction. |
Normal variance
Section titled “Normal variance”- Scores can move by one point on Q02 to Q05 between two runs on the same model, because the agent may pick a different reading of “latest month” or “unusually high” when the model doesn’t define them.
- On the AI-ready model the definitions are in the AI instructions and measures, so runs agree more often.
- The deterministic check is
Invoke-DaxQuestionTests.ps1in Lab 08. It runs the reference DAX against Desktop and doesn’t vary.