Turn 30 minutes of account research into 2

Create a workflow 'AI account deep research' triggered by account_created (high-tier accounts only — by ICP fit or domain-tier flag) OR manual trigger from a list view ('research these 50 accounts'). Goal: produce a structured account-intelligence brief that a rep can act on without doing their own research.

Configure web scrape step targeting the account's domain. Fetch: homepage hero copy, About page, Team/Leadership page, Pricing page (if exists), Customers page, recent blog posts. Returns raw page content to feed into the AI summarization step. Respect robots.txt; rate-limit per domain.

Configure AI research step that takes the scraped content and produces a structured summary: (a) one-sentence company description, (b) primary product/service, (c) ICP signals (target market, stated customer types), (d) recent priorities inferred from homepage / blog (e.g. 'expanding internationally', 'launching AI features'), (e) decision-maker names + titles from leadership page, (f) any mentioned competitors / partners, (g) confidence score on each finding.

Configure a second AI step that scores ICP fit using the company summary + a provided ICP definition (industry, size, tech, use cases). Output: numeric 0-100 score + reasoning string ('Strong fit because they are a B2B SaaS in [target industry] with ~50 employees and recently raised Series A — matches our ICP for growth-stage teams'). Records a structured fit_score + fit_reasoning attribute.

Configure AI write step that produces a draft outreach opener referencing specific findings (not generic). Format: 2-3 sentences referencing a SPECIFIC fact from the research ('I saw your team is expanding into APAC...'), then a value-hypothesis ('Teams growing internationally typically struggle with X — that's where we help'), then a soft CTA. Output stored on the account for SDR review-and-edit, never auto-sent.

Write back to the Account record: company_description, primary_product, recent_priorities, decision_makers_inferred (array), competitors_mentioned, fit_score, fit_reasoning, ai_drafted_opening, research_completed_at. The SDR sees a fully-formed account brief without doing any manual research.

Validate workflow DAG and publish. Add a daily summary digest of new accounts researched + their average fit scores so SDR managers can audit which accounts AI is flagging as best-fit.