1. Parse the request
   • Identify the target persona (see Persona Quick Reference below)
   • Extract company context (name, industry, size, any signals like funding, hiring, product launches)
   • Determine email type: first-touch or follow-up (default: first-touch)
   • If user provides a LinkedIn profile URL, proceed to step 1b

• Call

  mcp__crustdata__enrich_person_by_linkedin

  with the LinkedIn URL
• Extract and use the following for personalization:
  • Name and title: Use first name in greeting, title to identify persona
  • Current company: Company name, domain, description, funding stage
  • Career history: Past employers and roles (useful for "you've scaled teams before" angles)
  • Education: Schools and degrees (use sparingly, only if highly relevant)
  • Skills/languages: Can inform communication style
• Call

  mcp__crustdata__get_person_linkedin_posts

  to find recent posts for personalization hooks
• Call

  mcp__crustdata__enrich_company_by_domain

  with their company domain for deeper company intel (headcount, revenue, funding, founders)
• Personalization priorities from Crustdata data:
  1. Recent LinkedIn posts (best hook if they posted about relevant topic)
  2. Current role + company context (product-specific hooks)
  3. Career trajectory (e.g., "Since joining from [previous company]...")
  4. Company growth signals (funding, headcount growth)

1. Research the company first (CRITICAL for CS/CX leaders)
   • If Crustdata already provided company data, use it; otherwise use web search
   • Find the company's core products and platform names
   • Identify what their CS/support teams actually manage day-to-day
   • Look for product-specific terminology (e.g., "HealthRules Payer", "RingEX", "Qualtrics XM")
   • Find recent news, integrations, or platform updates
   • This research powers the subject line and hook
2. Load persona-specific guidance
   • Read

     references/personas.md

     for the matching persona archetype
   • Note their pain points, buying behavior, and anti-patterns
3. Match product capability to persona pain
   • Read

     references/product-intel.md

     for Inkeep product context
   • Identify which product pillar (Ask AI, Copilots, Workflows, Build Your Own) solves their problem
   • Select relevant proof point (e.g., "48% ticket reduction" for support, "18% activation" for product)
   • Never lead with product features, lead with outcome, then connect to capability
4. Select content CTA (optional but recommended)
   • Read

     references/blog-mapping.md

     to find relevant articles for this persona
   • Match buying stage: Awareness (cold), Consideration (exploring), Decision (evaluating)
   • For multi-step sequences: Select 2 articles with different angles for emails 2 and 3
5. Add social proof (when relevant)
   • Read

     references/customer-proof.md

     to find industry-matched customers
   • Use 1-2 customer names that match prospect by industry and size
6. Draft the email
   • Follow the Email Structure below
   • Apply persona-specific messaging angle
   • Weave in product benefit naturally (not as a pitch)
   • Keep it short (under 100 words for first-touch, under 150 for follow-ups)
7. Output the email
   • Provide subject line + body
   • If multiple variants requested, provide 2-3 options

https://www.linkedin.com/in/johndoe

• Name: John Doe
• Title: VP of Customer Success
• Company: Acme Corp (Series B, 200 employees, data analytics platform)
• Recent post: "Just shipped our new dashboard..."
• Past: Director of CS at DataCo

"Your team is proving ROI across Acme's analytics deployments while keeping pace with the dashboard updates you just shipped. Turning usage signals, support themes, and stakeholder feedback into renewal-ready briefs still takes manual stitching."

"I saw you lead Customer Success at Acme Corp." (wastes characters, no insight)

1. Fall back to web search for company research
2. Use title alone to identify persona archetype
3. Focus on company-specific hooks rather than personal hooks
4. Check if

   enrich_realtime: true

   option provides fresher data (costs more credits)

Subject: [2-3 words, internal-camo style, no punctuation]

[1 sentence: Personalized observation or trigger]

[1-2 sentences: Problem statement with loss framing or unconsidered need]

[1 sentence: Social proof:"We helped [similar company] [specific outcome] in [timeframe]"]

[1 sentence: Interest-based CTA with optional content offer]

• Under 100 words total (guideline, not hard rule)
• 3rd-5th grade reading level
• Plain text, no links in first-touch email
• 2-3 paragraphs, 1 sentence each
• Start with "you/your" not "I/we"
• Specific numbers, named companies, exact timelines
• Use hyphens only for compound words (edge-case, Tier-1, 50-80%). Never use dashes to connect separate thoughts or clauses.
• Never use em dashes. Use commas or periods instead.

[Company Product Name] QBR prep

• Qualtrics XM QBR prep

• Mural rollout QBR prep

• RingEX and RingCX QBR prep

• LeanIX onboarding, faster first value

• Vanta Trust Center QBR prep

• "With [Company] supporting [customer type] across [Product A] and [Product B], I imagine your team spends a lot of time pulling context together before QBRs and renewals."
• "For [product type] customers, the renewal story is usually scattered across [signal 1], [signal 2], and [signal 3]."
• "Turning [platform] activity, [metric], and [data source] into an exec-ready renewal story still takes hours of manual stitching."

• "I saw you lead Customer Success at [Company]." (Wastes characters on their title)
• "Your team probably spends time on manual work." (Not specific to their product)
• "I noticed you're hiring CSMs." (Generic trigger)

• Under ~80 words (executive micro emails)
• Peer-to-peer tone, not salesy or marketing copy
• Plain, credible language, avoid buzzwords
• Focus on leverage, capacity, and knowledge reuse, not features
• Assume they already have good tooling and possibly an AI assistant
• Personalize based on real company context

1. Customer-facing AI assistant with source-cited answers
2. Agent Copilot that analyzes requests and drafts replies with linked sources
3. Content writer that turns resolved tickets into KB/docs updates

"Inkeep gives you one shared knowledge layer that powers a cited customer AI assistant, an agent Copilot inside case workflows, and automatic capture of resolved cases into docs."

"A lot of support questions are already documented, but the right answer isn't surfaced in the moment, so cases still open and agents re-search."

"When a complex case closes, the resolution often stays in the case thread, while agents, customers, and AI tools keep using older guidance."

• Snowflake Copilot: keeping answers current

• Support leverage across 650+ integrations

  (New Relic)

• Clean Room setup: Snowflake + BigQuery

  (LiveRamp)

• SKAN + OneLink edge cases

  (AppsFlyer)

• When Qlik + Talend case learnings don't travel

• Docs exist. Finding them is the work

• When edge-case answers don't stick

• faster QBR prep

• Environment/version
• Configs
• Logs and traces
• Similar past cases
• Customer account state

• Focus on renewals, QBRs, and proving customer value
• Emphasize "bespoke CSM AI Agents tailored to your playbooks"
• Connect to CRM, ticketing, call notes, and product telemetry
• Three bullets: risk identification, health snapshot, exec-ready QBR/renewal brief

• [Company] renewals: [outcome metric], auto-summarized

• Before renewals: auto-brief on [outcome] + [outcome]

• [Company] CS: renewal risk + value story in 1 brief

• CS agent for [Company] renewals

• Smartly renewals: ROAS + creative velocity, auto-summarized

• BMC renewals: MTTR + automation wins, auto-briefed

• Before renewals: auto-brief on patch + exposure outcomes

  (Tanium)

• Nooks ROI brief: connects, meetings, pipeline

• Atlan: instant account intel for renewals

• Aqua renewals: measurable runtime impact, faster

• QBR-ready renewal brief for CSMs

• CS agent for renewals

• Select articles from

  references/blog-mapping.md

  that match the persona
• Email 2: Use article that adds new insight or reinforces problem framing
• Email 3: Use article with different angle (re-frame the problem)
• Include full URL on its own line for easy clicking
• Frame the article: "This covers [specific insight]:" then link
• Keep the ask soft: "Worth 5 minutes if [pain point] is on your radar"

• Email 2 has highest leverage (+49% reply lift)
• Follow-ups can be longer (4+ sentences get 15x more meetings)
• Avoid "I never heard back" (-14% meetings)
• "Hope all is well" works only when personalized to specific event
• Blog links add value without being salesy
• Never use meta-language like "Different angle:", "One stat that stood out:", "Bumping this", or "Here's another way to think about it:". Instead, just open with the new angle or stat directly. Let the content speak for itself.
• Use social proof only once per sequence (typically in Email 1). Repeating customer names across emails signals templated outreach.

1. Deep company research in Email 1: Referenced "800+ Snaps", "Groundplexes", "Ultra Tasks"
2. Progressive story arc: Knowledge trapped → Agent speed → Proactive support
3. Each email has ONE job: No repetition of the same pitch
4. Decreasing CTA intensity: Strong → question → soft → softer
5. All under 80 words: Respects exec time
6. No meta-language: Opens directly with the new angle

• Replace product data (800+ Snaps → their equivalent)
• Replace issue types (connector auth → their common escalations)
• Replace data sources (run logs, task status → their monitoring tools)

• Replace ticketing system (Zendesk → Salesforce, ServiceNow, etc.)
• Keep the "assist first, then automate" framing

• Usually works as-is since it's about the broader vision

• "Quick chat" / "Quick call" (trivializes their time)
• "Just following up" (no new value)
• Generic "I hope this finds you well"
• "We're a leading provider of..." (template smell)
• ROI claims without context (-15% success rate)
• Pitching your product first (-57% reply rate)
• Multiple CTAs (one ask per email)
• Wall of text (no paragraphs)
• Over 125 words on first touch
• Meta-language that signals templated outreach ("Different angle:", "One stat that stood out:", "Bumping this", "Circling back", "Following up on my last email", "Here's another way to think about it:")
• Sales-speak that reveals you're analyzing across prospects ("One pattern we see:", "What we're hearing from teams like yours", "A trend we've noticed")
• Em dashes anywhere in the email (use commas or periods instead)
• Opening with their title ("I saw you lead Customer Success at..." wastes characters and doesn't catch attention)
• Generic subject lines that apply to any company ("faster QBR prep" instead of "Qualtrics XM QBR prep")

• Starts with "I/My/We/Our"
• Contains buzzwords ("innovative", "cutting-edge", "all-in-one")
• Includes rounded numbers ("save 40%") instead of specific ("save 37%")
• Has generic social proof ("leading companies")
• Asks for meeting before establishing value
• Reads above 8th grade level
• Subject line could apply to any company (not product-specific)
• Opens with recipient's job title

• Show which tools are being accessed (signals real integrations)
• Collapse multiple steps into one readable action
• Sound proactive, not just reactive
• Add lightweight follow-ups ("Want next steps?") like real copilots
• Use clear business language ("primary champion", "risk factors", "expansion")
• Output feels exec-ready