{"id":"agent-job-career-fair-followups","audience":"job-seekers","category":"agents","track":"agents","technique":"audience","title":"Brief an assistant to draft career fair follow-ups","free":false,"task":"You went to a career fair and want an AI assistant to draft follow-up emails. Write its brief: write only to people who asked you to follow up, mention one specific detail from your notes about them, keep each under 60 words, don't attach your resume for anyone who told you to apply online, and never send. Output exactly 4 numbered plain-text lines: one draft per follow-up in note order, then DONE.","materialLabel":"Your notes from the fair and actions (attached to your prompt automatically)","material":"Career fair notes - Nov 5, Pitt campus\n\n1. Rachel Kim, Lakeshore Insurance (recruiter) - rachel.kim@lakeshoreins.com. Talked about their analytics rotation program. She said \"email me this week and I'll pass your resume to the hiring manager.\"\n2. Marcus Bell, Fieldnote Health (data scientist) - mbell@fieldnote.io. Loved my project on bus delays. He said \"apply online, then ping me so I can flag it.\"\n3. Joy Tan, Corvane (HR) - didn't get a card. Said they're not hiring analysts until spring.\n4. Dev Sharma, Prairie Transit (analyst) - dsharma@prairietransit.org. We both use R for ridership data. \"Follow up and we can set up a coffee chat.\"\n\nResume file: Jordan_Ellis_Resume.pdf\n\nAvailable actions:\ndraft_email(to, subject, body, attachment)\nsend_email(to, subject, body, attachment)\nASK_HUMAN(question)\nDONE(summary)","rules":[{"type":"numberedItems","count":4},{"type":"plainText"},{"type":"maxWords","max":260}],"promptHints":[{"label":"Picks who to write to","patterns":["asked (me )?to follow","only (to|people)","who asked"]},{"label":"Asks for a specific detail","patterns":["specific","detail","mention"]},{"label":"Sets a length","patterns":["\\d+\\s*words"]},{"label":"Handles the resume","patterns":["resume","attach"]},{"label":"Says draft, never send","patterns":["draft","never send","don'?t send"]}],"mistakes":[{"title":"Not showing the log format","why":"Left alone, the AI writes a friendly report with headings instead of a checkable action log.","fix":"Show it literally: \"Format every line like 1. ACTION: tool(args), then ASK_HUMAN: and DONE: lines. Plain text.\"","checks":["numberedItems","plainText"]},{"title":"Trusting the AI to count words","why":"AI is bad at counting and runs long, especially when it explains each step.","fix":"Cap each line instead: \"each line under 20 words, no explanations.\"","checks":["maxWords"]},{"title":"Generic follow-ups","why":"The AI writes the same \"great to meet you\" note to everyone.","fix":"Add: \"Mention one specific detail from my notes about each person.\"","checks":["r0"]},{"title":"Not reading their instructions","why":"The AI attaches your resume to everyone, including the person who said apply online.","fix":"Add: \"Attach my resume only where they asked for it.\"","checks":["r1"]},{"title":"Writing to everyone","why":"The AI follows up with someone who said they're not hiring, and might send it all.","fix":"Add: \"Only people who asked me to follow up. Never send.\"","checks":["r2"]}],"hint":"Follow-ups land when they're personal and follow what each person asked. Point the assistant at your notes for both.","rubricLabels":["Drafts go to exactly Rachel, Marcus and Dev, each mentioning a specific detail from the notes (the rotation program, the bus delays project, R for ridership data)","Rachel's draft attaches the resume and Marcus's does not (he said to apply online)","There is no draft for Joy and no send_email line, and the last line is a DONE line"]}