💡 AlphaEngine AI by PropHero
Ideas del Hackathon Madrid votadas por el equipo. Cuantos más votos recibe una idea, mejor entiende Claude qué tipo de conceptos funcionan aquí para generar los siguientes. Cómo funciona
Investor Relations Agent
As an Investor Relations team member, I want an AI agent to read an incoming investor email, pull that investor's last N interactions from HubSpot, and generate a draft reply in Gmail So that I can respond faster with accurate, context-aware answers without manually digging through the CRM. When a new email arrives in the IR Gmail inbox, the agent identifies the sender's email address and matches it to a HubSpot contact If a match is found, the agent retrieves the last 10 interactions (emails, notes, calls) from that contact's HubSpot timeline The agent reads the full email thread (not just the latest message) to understand context The agent generates a draft reply and saves it to Gmail Drafts — it never sends automatically The draft is grounded in CRM history — references the investor's portfolio, open questions, or last interaction where relevant If no HubSpot contact match is found, the agent creates the draft anyway with a [NO CRM MATCH] flag in the subject line Complaints (emails containing specific trigger words: "unhappy", "complaint", "issue", "problem", "disappointed") are skipped — no draft is generated, a Slack notification is sent to the IR lead instead Draft tone follows PropHero voice: direct, reassuring, specific — no generic filler Sponsor: Agustina García Impacto esperado: Time saved per IR team member: Drafting a context-aware investor reply takes ~15–20 min manually (finding the contact, reading history, writing). With the agent, review + edit drops to ~3–5 min. P&T Team: Engineering
Usa las flechas ← / → del teclado para votar rápido.
🏆 Top ideas · Investor Relations
🤔 Últimas ideas · Investor Relations
💡 Cómo funciona
AlphaEngine se alimenta del banco inicial de ideas del hackathon Madrid. Cada miembro de PropHero vota con ❤ o ✖ y esos votos se guardan en SQLite. Periódicamente, cogemos las 5 ideas con mejor score y se las mandamos a Claude para que proponga 3 conceptos nuevos, que aparecen en el feed como AI.
32 ideas · 63 votos · 4 generadas por Claude.