meeting-prep

Generates a one-page meeting prep brief before any sales or customer call. Pulls from HubSpot, Fireflies, and Apollo, cross-references data across sources, and delivers a structured brief with company

SKILL.md

GitHub

Meeting Prep — Pre-Call Intelligence Brief

Your job is to build a comprehensive, cross-referenced meeting prep brief before JP gets on a sales or customer call. Pull from every available data source, verify data consistency across sources, and deliver a structured brief that makes JP the most prepared person on the call.

Trigger Patterns

Activate this skill when the user says any of the following:

  • "prep for my meeting with [company/person]"
  • "meeting prep [company/person]"
  • "get ready for my call with [company/person]"
  • "prep me for [company/person]"
  • "what do I need to know before my call with [company/person]"
  • "brief me on [company/person]"
  • "prep [company/person]"
  • "call prep [company/person]"

Input Parsing

Extract the following from the user's request:

  • Company name (required) — the organization being met with
  • Person name (optional) — specific contact(s) at the company
  • Meeting date/time (optional) — when the meeting is scheduled

If only a person name is given, use HubSpot and Apollo to resolve their company. If ambiguous, ask the user to clarify before proceeding.

Data Collection Workflow

Step 1: Resolve Identifiers

Start by finding the company and contact in HubSpot so you have IDs for subsequent queries.

Call: mcp__claude_ai_HubSpot__search_crm_objects
Params:
  objectType: "companies"
  filterGroups: [
    { filters: [
      { propertyName: "name", operator: "CONTAINS_TOKEN", value: "<company name>" }
    ]}
  ]
  properties: ["name", "domain", "industry", "numberofemployees", "city", "state",
               "country", "phone", "description", "hs_lastmodifieddate",
               "notes_last_updated", "createdate", "hubspot_owner_id",
               "hs_num_open_deals", "hs_total_deal_value"]
  limit: 5

If a person name was provided:

Call: mcp__claude_ai_HubSpot__search_crm_objects
Params:
  objectType: "contacts"
  filterGroups: [
    { filters: [
      { propertyName: "firstname", operator: "CONTAINS_TOKEN", value: "<first name>" },
      { propertyName: "lastname", operator: "CONTAINS_TOKEN", value: "<last name>" }
    ]}
  ]
  properties: ["firstname", "lastname", "email", "jobtitle", "phone",
               "company", "hs_lead_status", "lifecyclestage", "createdate",
               "notes_last_updated", "hs_sales_email_last_replied",
               "hs_email_last_reply_date", "hubspot_owner_id",
               "hs_sequences_is_enrolled"]
  limit: 5

If no person name was provided, search for all contacts associated with the company.

Step 2: Pull All Data Sources (run in parallel)

2A. HubSpot — Deals

Call: mcp__claude_ai_HubSpot__search_crm_objects
Params:
  objectType: "deals"
  filterGroups: [
    { filters: [
      { propertyName: "associations.company", operator: "EQ", value: "<companyId>" }
    ]}
  ]
  properties: ["dealname", "amount", "dealstage", "pipeline", "closedate",
               "createdate", "hs_lastmodifieddate", "hubspot_owner_id",
               "hs_deal_stage_probability", "description",
               "notes_last_updated", "hs_is_closed_won",
               "hs_is_closed", "hs_acv", "hs_mrr"]
  limit: 20

If company association search does not work, search by deal name containing the company name as a fallback.

Also get deal stage names:

Call: mcp__claude_ai_HubSpot__get_properties
Params: objectType="deals", propertyNames=["dealstage", "pipeline"]

2B. HubSpot — Engagement History

Search for contacts associated with the company to understand the full engagement timeline:

Call: mcp__claude_ai_HubSpot__search_crm_objects
Params:
  objectType: "contacts"
  filterGroups: [
    { filters: [
      { propertyName: "company", operator: "CONTAINS_TOKEN", value: "<company name>" }
    ]}
  ]
  properties: ["firstname", "lastname", "email", "jobtitle", "phone",
               "hs_lead_status", "lifecyclestage", "createdate",
               "notes_last_updated", "hs_sales_email_last_replied",
               "hs_email_last_reply_date", "hs_email_sends_since_last_engagement",
               "num_notes", "num_contacted_notes", "hs_sequences_is_enrolled",
               "hs_analytics_num_page_views", "hs_analytics_num_visits"]
  limit: 20

2C. Fireflies — Past Meeting Transcripts

Search for meetings with this company/person:

Call: mcp__claude_ai_Fireflies__fireflies_search
Params:
  keyword: "<company name>"
  limit: 10

Also try searching by person name if provided:

Call: mcp__claude_ai_Fireflies__fireflies_search
Params:
  keyword: "<person name>"
  limit: 10

For each relevant meeting found, pull the summary:

Call: mcp__claude_ai_Fireflies__fireflies_get_summary
Params: transcriptId: "<meeting ID>"

Extract from each meeting:

  • Key discussion topics and decisions
  • Pain points mentioned by the prospect/customer
  • Pricing or budget discussions (exact numbers)
  • Action items assigned to either party
  • Objections raised
  • Competitors mentioned
  • Timeline or urgency signals
  • Direct quotes that reveal priorities or concerns

2D. Apollo — Company Enrichment

Call: mcp__claude_ai_Apollo_io__apollo_organizations_enrich
Params:
  domain: "<company domain from HubSpot>"

Extract:

  • Company size (headcount, headcount growth)
  • Funding information (total raised, last round, investors)
  • Industry and sub-industry
  • Technology stack
  • LinkedIn URL
  • Key executives
  • Annual revenue estimate
  • Recent news or job postings

If domain is not available, search by name:

Call: mcp__claude_ai_Apollo_io__apollo_mixed_companies_search
Params:
  q_organization_name: "<company name>"
  per_page: 3

2E. Apollo — Contact Enrichment

If a specific person is the meeting attendee:

Call: mcp__claude_ai_Apollo_io__apollo_people_match
Params:
  email: "<contact email>"

Or if no email:

Call: mcp__claude_ai_Apollo_io__apollo_mixed_people_api_search
Params:
  q_person_name: "<person name>"
  q_organization_name: "<company name>"
  per_page: 3

Extract:

  • Current title and seniority
  • LinkedIn profile URL
  • Previous companies and roles
  • Time in current role
  • Department

2F. Apollo — Job Postings (Buying Signals)

Call: mcp__claude_ai_Apollo_io__apollo_organizations_job_postings
Params:
  organization_id: "<apollo org ID>"

Look for:

  • Hiring for roles related to Solum's value prop (billing, RCM, operations, tech)
  • Growth signals (lots of clinical hiring = scaling)
  • Pain signals (hiring for roles Solum could replace or augment)

Step 3: Internal Fact-Check Layer

Before assembling the brief, cross-reference data across sources. Check for and flag:

Data Consistency

  • Title mismatch: Does the contact's title in HubSpot match Apollo? If not, note which is likely current.
  • Company size discrepancy: Does HubSpot's employee count match Apollo's? Flag if >20% difference.
  • Deal amount vs. transcript: If Fireflies transcripts mention specific pricing ($X/mo, $X/yr), compare against the HubSpot deal amount. Flag any discrepancy with exact quotes.
  • Contact info: Is the email/phone in HubSpot consistent with Apollo?

Staleness Detection

  • Deal stage age: Calculate days in current stage. Flag if >30 days with a warning.
  • Last activity gap: Calculate days since last touchpoint (email, call, meeting, note). Flag if >14 days.
  • Close date accuracy: Is the close date in the past? Has it been pushed more than once?
  • Stale contact data: If HubSpot contact was last modified >90 days ago, note the data may be outdated.

Action Item Tracking

  • Review action items from previous Fireflies meetings.
  • Cross-reference against HubSpot notes and activity timeline.
  • Classify each action item as:
    • COMPLETED: Evidence of follow-through in HubSpot activity or subsequent meeting
    • PENDING: No evidence of completion
    • UNKNOWN: Cannot determine status from available data

Missing Data Flags

  • No deal associated with this company? Flag it.
  • No email history? Flag it.
  • No previous meetings? Note this is a first meeting.
  • Contact has no title? Flag for update.

Step 4: Assemble the Brief

Present the brief in this exact format. Keep it dense and scannable. No filler.

================================================================
  MEETING PREP: [Company Name]
  [Meeting Date if known] | Prepared [today's date]
================================================================

ATTENDEES
---------
[Name] — [Title] (from [source: HubSpot/Apollo])
[Name] — [Title]
  LinkedIn: [URL if available]
  Email: [email] | Phone: [phone]
  Time in role: [X months/years, from Apollo]
  Previous: [previous company/role if notable]

COMPANY SNAPSHOT
-----------------
Industry:      [industry]
Size:          [employee count] ([growth trend if available])
Location:      [city, state]
Founded:       [year]
Funding:       [total raised / last round / investors]
Revenue est:   [if available from Apollo]
EMR/Tech:      [technology stack from Apollo]
Domain:        [website]
LinkedIn:      [company LinkedIn URL]

Key facts:
- [Notable fact 1 from Apollo enrichment]
- [Notable fact 2]
- [Recent job postings relevant to Solum]

RELATIONSHIP HISTORY
---------------------
First contact:     [date from HubSpot createdate]
Total touchpoints: [count of emails + calls + meetings + notes]
Last interaction:  [date and type — email/call/meeting]
Days since last:   [calculated]

Timeline:
  [Date] — [Event: first email, first call, demo, proposal sent, etc.]
  [Date] — [Event]
  [Date] — [Event]
  ...

DEAL STATUS
------------
Deal:          [deal name]
Stage:         [current stage name] (since [date])
Days in stage: [calculated] [FLAG if >30 days]
Amount:        $[amount] ([monthly/annual])
Close date:    [date] [FLAG if past due or pushed]
Pipeline:      [pipeline name]
Probability:   [%]
Owner:         [name]

[If multiple deals, list each]
[If no deal exists: "No deal record found — consider creating one after this meeting"]

PREVIOUS MEETING HIGHLIGHTS
-----------------------------
[Meeting date] — [meeting title]
  Key topics discussed:
  - [topic 1]
  - [topic 2]

  Their pain points (in their words):
  - "[direct quote from transcript]"
  - "[direct quote]"

  Decisions made:
  - [decision 1]

  Competitors mentioned:
  - [competitor name + context]

  Action items from this meeting:
  - [JP] [action item] — Status: [COMPLETED/PENDING/UNKNOWN]
  - [Prospect] [action item] — Status: [COMPLETED/PENDING/UNKNOWN]

[Repeat for each past meeting, most recent first]
[If no meetings found: "No previous meetings on record. This appears to be a first meeting."]

THEIR PRIORITIES (from their own words)
-----------------------------------------
Based on transcript analysis and email history:
1. [Priority 1 — with source: "quote" from meeting on date]
2. [Priority 2 — with source]
3. [Priority 3 — with source]

[If no transcript data: "No direct quotes available. Priorities inferred from deal context and industry."]

OPEN QUESTIONS / RISKS
------------------------
- [Unresolved question from past meetings]
- [Risk factor: e.g., "No executive sponsor identified"]
- [Risk factor: e.g., "Close date has been pushed twice"]
- [Risk factor: e.g., "Competitor [X] was mentioned favorably in last call"]
- [Missing info: "Budget not discussed yet"]
- [Missing info: "Decision-making process unclear"]

RECOMMENDED TALKING POINTS
----------------------------
Lead with:
1. [Specific opener based on their priorities and last interaction]
2. [Reference to specific pain point they mentioned]

Address:
3. [Pending action item or follow-up from last meeting]
4. [Objection handling for known concerns]

Advance the deal:
5. [Specific ask to move to next stage]
6. [Timeline or urgency angle based on their signals]

Avoid:
- [Topic or approach to stay away from, with reason]
- [Sensitive area based on past interactions]

FACT-CHECK NOTES
-----------------
[List any discrepancies found during cross-referencing]
- [e.g., "Title mismatch: HubSpot says 'Director of Ops', Apollo says 'VP Operations' — Apollo updated more recently, likely current"]
- [e.g., "Deal amount is $50K in HubSpot but transcript from Feb 12 discussed $4K/month ($48K/yr) — minor discrepancy, verify on call"]
- [e.g., "Action item from Jan meeting (send case study) — no evidence of completion in HubSpot"]

[If no discrepancies: "All data consistent across sources."]

DATA FRESHNESS
---------------
HubSpot company: last modified [date]
HubSpot contacts: last modified [date]
HubSpot deal: last modified [date]
Fireflies: [X] meetings found, most recent [date]
Apollo: enrichment as of [today]
================================================================

Error Handling

  • If HubSpot returns no company match: try alternate spellings, abbreviations, or domain search. If still nothing, note "Company not found in HubSpot" and continue with Apollo data.
  • If Fireflies returns no meetings: say "No previous meetings recorded." Continue with other sources.
  • If Apollo enrichment fails: skip the company snapshot details that require Apollo. Note the source was unavailable.
  • If any MCP tool fails: continue with available data. Note which sources were unavailable at the top of the brief.
  • Never skip the brief because one source failed. Deliver what you have, clearly noting gaps.

Tone & Style

  • Dense, scannable, no fluff. This is a working document, not a narrative.
  • Data-first. Lead with facts, not opinions.
  • Direct quotes from transcripts are gold. Always include them when available.
  • Every recommendation must be grounded in data from the sources. No generic sales advice.
  • Flag unknowns explicitly. "Unknown" is better than a guess.
  • The entire brief should be readable in under 3 minutes.

Post-Brief Actions

After delivering the brief, ask:

  1. "Want me to pull the full transcript from any of these meetings?"
  2. "Should I update any stale data in HubSpot before your call?"
  3. "Need me to draft any follow-up materials for after the meeting?"