Daily prospecting dashboard with buying signals and automated outreach drafts
Build a daily prospecting workspace for an Account Executive at Tines (a security/IT workflow automation platform). Its job is to help an AE pick a prospecting mode, pull the right accounts and buying signals, identify the correct contacts, and draft personalized, signal-led outreach — while keeping a human in the loop for every send. Ship it in two parts: a persistent dashboard app and a chat-driven assistant, both backed by the same store. Persist all state in one datastore owned by the workflow (a SQLite database on a workflow volume works well — or your preferred persistence). Store accounts, buying signals, person-level hiring/leadership changes, contacts with a verified-email flag, email drafts, a sent/outreach log, and a single "standing standards" record (ICP definition, tone rules, do-not-contact list, and signal-priority order that the assistant reuses each session instead of re-asking). Seed it with realistic mock data so everything is usable before real sources are connected, and design the schema so live data can be swapped in later without changing the UI or agents. Triggers and main steps: A seed step, run manually once, that creates the schema and loads sample data. A JSON data API (HTTP route, space-authenticated) that the dashboard reads from and writes to — listing signals, hiring/leadership moves, accounts, coverage, drafts, and settings, plus mutations to rate signals, manage/approve/send drafts, and edit settings. A dashboard web app (HTTP route) as the entry point: a tabbed single-page app with Signals, Hiring & Leadership, Target Accounts, Coverage, Drafts/Outreach, an Assistant chat tab, and Settings. Include global account search, a row-click account detail view that links a company's signals/contacts/drafts together, a manual refresh, and a per-row "Draft email" action. A draft-email AI step (HTTP route, called by the dashboard's action) that picks the best contact and trigger signal for one account and writes a short, signal-led email — but holds rather than drafting if there's no real signal or no verified email. A conversational prospecting assistant (an AI agent behind an HTTP route, driving the chat tab) that runs the full flow: ask the mode (fresh ICP list / buying signals / intent data / an existing list), pull a ranked account list, find contacts, draft outreach into the queue, and give a compact recap. It reads the standing standards and never contacts anyone on the do-not-contact list. A scheduled refresh (cron, e.g. daily) that tops up the signal feed and is the designated insertion point for real data sources. Use an LLM for the drafting step and the assistant (I used Claude via Anthropic — or your preferred model provider). For the prospecting data, wire in a signals/firmographics source (I planned Sumble — or your data provider), an intent source (I planned 6sense — or your intent platform), a CRM for pipeline and do-not-contact (Salesforce/HubSpot — or your CRM), and a contact-enrichment and GTM-data source (Clay for its strong enrichment/waterfall and go-to-market capabilities, or Apollo/Clearbit — or your enrichment tool). Keep these behind the same datastore tables so they're a clean swap, and start on mock data. Keep all routes private to the space unless you decide otherwise. Emails should lead with the specific trigger, stay under ~100 words, avoid clichéd openers, and never be sent automatically — a human approves each one, and marking a draft "sent" updates the account's last-touch date feeding the Coverage view. Tools used: 3B workflow volume, SQLite, React/Tailwind dashboard, Anthropic Claude API, Sumble, 6sense, Clay, CRM (Salesforce/HubSpot), contact enrichment (Apollo/Clearbit)
What this prompt builds
A daily prospecting workspace for account executives that monitors buying signals across firmographic, intent, and hiring data sources, identifies in-market accounts with trigger events, enriches contacts weighted toward role changes, and drafts personalized, signal-led outreach emails. The system maintains a persistent dashboard with tabs for signals, leadership changes, target accounts, coverage, and drafts, plus a conversational assistant that orchestrates the full prospecting flow while enforcing ICP rules and do-not-contact lists. All state persists in a workflow datastore so every signal, contact, and draft remains browsable and actionable rather than disappearing after one session.
The problem
Account executives waste hours each week hunting for which accounts are actually in-market right now. Buying signals — funding rounds, leadership changes, tech stack shifts, hiring spikes — exist scattered across intent platforms, firmographic tools, and enrichment sources, but nobody has time to check them daily, cross-reference them against a target list, find the right contact, and turn a hot signal into a personalized email before it goes cold or a competitor gets there first. Teams that rely on manual prospecting lose deals to timing, miss high-reply-rate moments like role changes, and let strong signals expire in tabs they never revisit.
Solution and impact
Account executives get a persistent dashboard and conversational assistant that automatically pull ranked, in-market accounts with a specific "why now" trigger attached, identify the best contact weighted toward people who just changed roles, and draft short, signal-led emails ready for human approval. The workflow maintains live tabs for signals, hiring and leadership changes, target account coverage, and draft outreach so nothing disappears after one session, and it enforces ICP definitions and do-not-contact rules so every email is relevant and compliant. Teams close more deals by reaching the right person at the exact moment a buying signal appears, and AEs reclaim hours previously spent on manual research and list-building.