El Prompt de Triage de Bandeja de Entrada: Convierte más de 100 correos no leídos en una lista de acciones clasificada en una sola pasada
Why this prompt matters
Knowledge workers spend roughly 11 hours a week managing email, and heavy users lose closer to 15. Nearly 11 of those hours go to messages that didn't need real-time attention at all. The real cost isn't the time spent reading — it's the urgent message buried on page 3 that gets answered two days too late because it looked identical to 40 other messages in a flat, unsorted inbox.
What we use it for
You're back from a 3-day conference with 140 unread emails and 45 minutes before your next meeting. Somewhere in that pile is a client threatening to cancel, a deadline moved up, and a dozen threads that can wait until next week — but you can't tell which is which without reading every single one.
Prompt
You are an executive assistant with 15 years of experience triaging high-volume inboxes for busy professionals. You are precise, fast, and never bury the lede.
CONTEXT:
I am a [YOUR ROLE, e.g. "product marketing manager"] returning to [NUMBER] unread emails after [TIME PERIOD, e.g. "a 3-day conference"]. I have [AVAILABLE TIME, e.g. "45 minutes"] before my next commitment. My current priorities are: [LIST YOUR TOP 2-3 PRIORITIES, e.g. "closing the Q3 renewal with Client X, prepping the board deck due Friday, not missing anything from my direct reports"].
TASK:
Below is a dump of my unread emails (sender, subject line, and first 2-3 sentences of each). For EACH email:
1. Assign an Urgency score (1-5): how time-sensitive is this, based on explicit deadlines, escalation language, or repeated follow-ups.
2. Assign an Impact score (1-5): how much does this affect my stated priorities or key relationships if ignored.
3. Calculate a Priority Score (Urgency x Impact).
4. Sort ALL emails into four action buckets: RESPOND NOW (today), RESPOND SOON (this week), DELEGATE OR SCHEDULE, and ARCHIVE OR IGNORE.
5. For every email in RESPOND NOW, draft a 2-4 sentence reply I can send with minimal editing.
CONSTRAINTS:
- Do not summarize an email's full content — I've already seen the subject line, I need your JUDGMENT on priority, not a restatement.
- If two emails from the same sender are duplicates or follow-ups on the same thread, merge them into a single line item and note the follow-up count.
- Flag anything that mentions a deadline, a dollar amount, or an escalation phrase ("following up again", "need this by", "per our conversation") with a ⚠️ marker regardless of your calculated score.
- If information is missing to judge urgency or impact, ask me ONE clarifying question about that specific email rather than guessing.
OUTPUT FORMAT:
## RESPOND NOW (today)
[Sender] — [Subject] — Priority: [score] ⚠️ if flagged
Draft reply: [2-4 sentence draft]
## RESPOND SOON (this week)
[Sender] — [Subject] — Priority: [score] — Why it can wait: [one line]
## DELEGATE OR SCHEDULE
[Sender] — [Subject] — Suggested owner or calendar action
## ARCHIVE OR IGNORE
[Sender] — [Subject] — one-line reason
EMAILS TO TRIAGE:
[PASTE YOUR EMAIL DUMP HERE — sender, subject, and first few lines of each]Result
RESPOND NOW (today)
Sarah Chen (Client X, VP Procurement) — "Following up again on renewal terms" — Priority: 20 ⚠️ (3rd follow-up, contract expires in 6 days)
Draft reply: "Hi Sarah, apologies for the delay — I've been at a conference this week. I have your revised terms in front of me now and will have a redlined response back to you by end of day tomorrow. Can we lock in a 15-minute call Thursday morning to close this out?"
Marcus Webb (Board Secretary) — "Board deck: need final slides by Thursday 5pm" — Priority: 16 ⚠️ (hard deadline moved up from Friday)
Draft reply: "Thanks for the heads up on the moved deadline. I'll have the final revenue and roadmap slides to you by Wednesday EOD so there's buffer for review. Flagging now in case the timeline shifts further."
RESPOND SOON (this week)
Priya Patel (Direct Report) — "Need sign-off on Q3 hiring plan" — Priority: 12 — Why it can wait: no stated deadline, but affects team planning; address after board deck is locked.
David Kim (Partner, Acme Co) — "Quick question about the API integration timeline" — Priority: 9 — Why it can wait: informational, no urgency language detected.
DELEGATE OR SCHEDULE
IT Helpdesk — "Your laptop warranty expires soon" — Suggested owner: forward to office manager.
Newsletter: Industry Weekly — Suggested action: schedule 10-min read for Friday, not urgent.
ARCHIVE OR IGNORE
LinkedIn notifications (4 threads) — No action needed, batch archive.
Internal all-hands recap — Informational only, no response required.
⚠️ One clarifying question: the email from "Legal — NDA Review" doesn't specify a deadline or which deal it relates to. Can you confirm which contract this ties to so I can score it accurately?
La mayoría de los prompts de gestión de bandeja de entrada hacen una de dos cosas: resumir todo (inútil — ya viste las líneas de asunto) o pedirte que describas tus prioridades en términos vagos y esperar que la AI adivine correctamente. Este Prompt no hace ninguna de las dos. Impone una rúbrica de puntuación — Urgencia por Impacto — para que la AI tenga que justificar cada clasificación en lugar de hacer coincidir patrones con palabras clave como "urgente" en la línea de asunto.
Por qué importa la rúbrica de puntuación
Separar la Urgencia del Impacto captura una categoría de correo que los prompts de coincidencia de palabras clave pasan por alto por completo: el mensaje silenciosamente importante. El tercer seguimiento de un cliente sobre una renovación de contrato no grita urgencia en su línea de asunto, pero el patrón de seguimiento repetido y la cantidad de dinero en juego lo hacen más importante que un correo en mayúsculas "URGENTE" de un boletín. El Prompt instruye explícitamente al modelo a señalar lenguaje de escalamiento, plazos y cantidades de dinero independientemente de la puntuación calculada — una red de seguridad contra la posibilidad de que la propia rúbrica se pierda algo.
Por qué hace una pregunta aclaratoria en lugar de adivinar
La mayoría de los prompts de un solo disparo fuerzan al modelo a adivinar cuando falta información, lo que silenciosamente degrada la precisión en los correos que más importan — los ambiguos. Instruir al modelo para que haga UNA pregunta aclaratoria cuando no puede juzgar urgencia o impacto mantiene la salida confiable en lugar de estar confiadamente equivocada. Respondes la pregunta, vuelves a ejecutar el triage y obtienes una clasificación corregida en lugar de descubrir que la AI juzgó mal las consecuencias de un contrato después del hecho.
Por qué se redactan respuestas solo para elementos RESPOND NOW
Redactar respuestas para los 140 correos sería ruido — la mayoría no necesita una respuesta meditada, sino una categoría. Limitar las respuestas redactadas al nivel RESPOND NOW significa que obtienes el resultado de mayor apalancamiento exactamente donde lo necesitas: el puñado de mensajes donde una respuesta rápida y bien redactada realmente cambia un resultado hoy.
Cómo adaptarlo
Intercambia tu rol real, lista de prioridades y tiempo disponible antes de cada ejecución — el modelo pondera todo en función de esas tres entradas, por lo que entradas vagas producen clasificaciones vagas. Para uso recurrente, guarda tu lista de prioridades como una nota que pegues cada vez en lugar de reescribirla, y actualízala semanalmente a medida que tus prioridades cambien.