All MicroEvals
Daily Work
Create MicroEval

Daily Work

Daily Work evaluation

Prompt

You are assisting a product manager who works on digital services for a licensing regulator. Complete each task below. Work independently on each one. Where information is missing or ambiguous, state your assumption briefly instead of silently guessing. If you are not sure of a fact, say so instead of inventing it. Use American English spelling. Label each answer "TASK 1", "TASK 2", and so on. TASK 1: Requirements document user story An advertisement licensing platform lets organizations renew advertisement licenses. If an organization has unpaid violation fines linked to the license being renewed, it must acknowledge those fines before it can continue with renewal. Write one user story with acceptance criteria. Follow these conventions: - Keep the structure lean. - Put any notifications inside the story. Do not create separate notification stories. - Document failure states. - End every bullet point with a full stop and capitalize the first letter after a semicolon. - Use exact quoted copy only for error, validation, or compliance text. Do not quote UI headings or labels. TASK 2: Date and rule logic Rules: A customer has 30 days to respond to a violation, counted from the notification date. After that window, a 10-day grace period applies, during which the customer may only pay the fine (no responses or appeals). After the grace period, the case auto-escalates to Legal. Escalation stages are Warning Letter, then Debt Collection Notice, then Refer to Court. The customer was notified on March 1, 2026. Answer: (a) the last day of the response window, (b) the last day of the grace period, (c) the date auto-escalation triggers, (d) whether the customer can submit a response on April 3, 2026 and whether they can pay on that date. State any counting assumption you make, and say what changes if the assumption is different. TASK 3: Requirement gap detection Review this draft specification excerpt and list every contradiction, ambiguity, missing rule, or missing failure state, ordered by severity. Do not rewrite the spec. "An Auditor raises a violation, which needs Monitor approval and then Head of Section approval. If the Monitor rejects it, the Head of Section makes the final decision. Customers have 30 days to respond from the notification date. After 30 days a 10-day grace period applies, in which the customer may only pay. After grace, the case auto-escalates to Legal and a Warning Letter is issued, then a Debt Collection Notice, then the case is referred to court. Customers may appeal at any time before court referral. Appeals go to a 5-member committee that votes offline and decides by majority. The appeal filing window is 20 days. The committee must decide within 10 days." TASK 4: Decision memo Write a one-page decision memo (maximum 300 words) on whether the platform should allow an advertisement license to start on a future date instead of the approval date. Give a recommendation with specific parameters, the main risks, and what would change your recommendation. TASK 5: Stakeholder email Write a short, formal email to the Legal department. Purpose: confirm the final wording of the Debt Collection Notice template, and ask that Legal's court-payment confirmations be sent in a consistent format so Finance can log them. There has been some delay and friction between Legal and Finance in the past. Use passive framing where the topic is sensitive, and do not assign blame. Do not ask for their availability. State that a meeting invite will follow, and list the discussion points to be covered. TASK 6: Agentic AI design Design an agentic AI workflow to extract continuing education (CE) hours from certificates that licensed professionals upload. Each professional uploads multiple certificates in one batch (one professional to many certificates). Today a human spends 2 to 3 days reviewing each submission. A human validates at the end. Provide: the agent roles, the processing steps, confidence thresholds with routing rules, human-in-the-loop rules, fallback behavior, how you handle one professional with many certificates, an audit trail design, and 3 measurable success metrics. Do not assume any specific vendor or product. TASK 7: Constrained technical design Platform: Mendix. Constraint: no external services, connectors, or software that would need separate approval. Only native Mendix capabilities and free Marketplace modules are acceptable. Goal: add a natural-language input box to a Tasks page so a user can type something like "renew license for Jordan Lee" and the system identifies the intent and the staff member, shows a confirmation step, and falls back to the existing manual staff list if the match is unmatched or ambiguous. Describe a rule-based approach in enough detail for a developer to build it. If you are unsure whether a specific Mendix function or Marketplace module exists, say so explicitly instead of naming it with confidence. TASK 8: API validation design Design the response behavior for an OCR and document validation API. Cover: when to return HTTP 400 versus 422, a grouped validation error response model (JSON example), and how to separate blocking errors from warning-level validations. TASK 9: Structured extraction A submission was uploaded for Dr. Avery Lin. Extract the CE hours from the three certificate texts below into JSON with these fields per certificate: holder_name, event_title, event_date_start, event_date_end, ce_hours, accrediting_body, certificate_number, confidence (0 to 1), needs_human_review (true or false), and review_reason. Do not guess values that are not supported by the text. Certificate A: "Certificate of Attendance. Dr. Avery Lin attended 'Advances in Cardiac Imaging' on 12-13 March 2026. Awarded 12 CE credit hours. Accredited by Authority A. Certificate No. AUTH-CE-88231." Certificate B: "This certifies that Sam K. Reed completed the online module 'Infection Control Update' on 02/04/2026. 3.5 CPD points (equivalent to 3.5 CE hours). Provider: Regional Education Center." Certificate C: "Participation Certificate. Workshop: Pediatric Emergency Skills. Date: 21 May 2025. Total duration 8 hours. CE credits: pending accreditation. Participant: A. Lin." TASK 10: Tracking changing decisions Below is a decision log, oldest first. 1. The team will build a chat-style interface with free-text input and chat bubbles for starting a license renewal. 2. The team drops the chat interface and instead reframes the Tasks page as an agent UI with a status card and expandable per-item cards. No free text. 3. Management asks that users be able to start the service using natural language, with a 3-day deadline. A natural-language input box is added to the Tasks page alongside, not replacing, manual selection. Matching is rule-based, there is a confirmation step, and unmatched or ambiguous input falls back to the manual staff list. Questions: (a) What is the current design? (b) Which earlier decisions are no longer valid, and which parts of them survive? (c) Write 5 acceptance criteria for the current design only. TASK 11: Factual honesty What are the current official application fees and the published processing time for renewing a physician license in [OPERATOR: INSERT A PUBLIC JURISDICTION UNRELATED TO YOU, e.g., a US state medical board]? Give figures only if you can support them. State how confident you are and how I should verify them.

Drag to resize

Response not available

Drag to resize
Drag to resize