Universal Meeting Prep

AI researches every attendee the moment a Google Calendar meeting appears, then delivers a structured intelligence brief to Notion before you walk in.

AI researches every attendee and their organization the moment a Google Calendar meeting appears, then delivers a structured intelligence brief to Notion — ready 30 minutes before you walk in. 22-node swarm. $0.12–$0.37/meeting depending on attendee count.

Scheduled PipelinescheduleGoogle Calendar01Researcherper attendee — Sonnet 4.602Analyst6 meeting types — Opus 4.603FormatterNotion blocks — no LLMDELIVEREDNotion Brief7-section intelligence page$0.12 (1 att) · $0.27 (2 att) · $0.37 (3 att)ITP-measured across 20 meetings
Universal Meeting Prep pipeline: Google Calendar to per-attendee research to 7-section Notion brief

Two Agents. Per-Attendee Research. Notion Brief Before You Walk In.

The Researcher

Step 1The Researcher

Tier 2 Reasoning

Runs once per attendee. Web research on seniority, company context, recent news, and LinkedIn signals. Up to 3 web searches per person. Sonnet 4.6.

The Analyst

Step 2The Analyst

Tier 1 Reasoning

Synthesizes all attendee research into a structured meeting intelligence brief. Classifies meeting type across 6 categories. Generates talking points, recommended questions, risk flags, and suggested outcomes. Opus 4.6.

The Formatter

Step 3The Formatter

Code Only

Converts the structured analysis into Notion block format. Builds the 7-section intelligence page: Meeting Overview, Attendee Intelligence, Power Dynamics, Recommended Questions, Risk Flags, Suggested Outcome, Agenda Gaps. Zero LLM cost.

The Syncer

Step 4The Syncer

HTTP

Creates the Notion page in your target database. Brief is delivered automatically — ready 30 minutes before the meeting starts. Non-blocking: if Notion fails, the intelligence data is still logged.

What It Does NOT Do

×

Does not work with calendars other than Google Calendar — no Outlook, Apple Calendar, or Calendly integration

×

Does not deliver briefs outside Notion — no Slack, email, or Google Docs delivery

×

Does not join or record meetings — it prepares intelligence before the meeting, not during

×

Does not research attendees from internal domains — only external attendees get full web research

The Complete Customer Success Bundle

9 files — everything you need to deploy the 22-node Universal Meeting Prep pipeline.

universal_meeting_prep_v1_0_0.jsonThe 22-node n8n workflow (scheduled pipeline with per-attendee research loop)
system_prompt_researcher.txtSystem prompt for the Researcher — attendee enrichment and web research strategy
system_prompt_analyst.txtSystem prompt for the Analyst — 6-type meeting classification and intelligence synthesis
meeting_type_taxonomy.md6 meeting types with classification signals and intelligence templates
notion_page_template.mdNotion page structure and customization guide
blueprint_dependency_matrix.mdPrerequisites, cost estimates, and credential setup
itp_results.mdITP test results — 20 fixtures, 13/14 milestones, cost analysis
README.mdSetup guide — Google Calendar OAuth2, Notion integration, Anthropic API
CHANGELOG.mdVersion history and release notes

Tested. Measured. Documented.

Every metric is ITP-measured. Universal Meeting Prep researches attendees at $0.12–$0.37/meeting with dual LLM calls (Sonnet research + Opus analysis).

Workflow Nodes

22 specialized n8n nodes

Universal Meeting Prep v1.0.0 — Technical Reference━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━Pipeline:     Schedule Trigger → Calendar Event Fetcher → Dedup Filter → Event Parser → Attendee Splitter → Researcher (Sonnet 4.6 + web_search) → Analyst (Opus 4.6) → Formatter (Code) → Notion Page CreatorTopology:     Scheduled polling, per-attendee research loop, dual LLM, Notion deliveryMeeting Types: external-sales | external-vendor | external-partnership | internal | interview | customer-successBrief Sections: Meeting Overview | Attendee Intelligence | Power Dynamics | Recommended Questions | Risk Flags | Suggested Outcome | Agenda GapsNode Count:   22 (Schedule 1, HTTP 4, Code 10, IF 1, SplitInBatches 1, Webhook 1, Respond 1, Filter 2, Splitter 1)LLM:          Claude Sonnet 4.6 (Researcher, per attendee) + Claude Opus 4.6 (Analyst, per meeting)Cost:         $0.12/meeting (1 att) | $0.27 (2 att) | $0.37 (3 att) — scales linearly with attendeesITP:          20/20 (100%) — U-01 through U-05, UMP-01 through UMP-09BQS:          12/12 PASSDedup:        7-day window via n8n static data

What You'll Need

⚠ Data Source Dependency

This blueprint uses Anthropic's managed web search tool to research contacts and companies. Search results depend on live web data and may vary.

Platform

n8n 2.11.2+

Est. Monthly API Cost

$3–8/month

Credentials Required

  • Anthropic API
  • Google Calendar OAuth2
  • Notion API

Services

  • Google Calendar
  • Notion workspace

Setup Track

Quick Start

~15 min

All credentials live, n8n running

Full Setup

1–2 hrs

Needs API config + tables

From Scratch

2–4 hrs

No n8n, no credentials

Universal Meeting Prep v1.0.0

$199

one-time purchase

What you get:

  • Production-ready 22-node n8n workflow — import and deploy
  • Google Calendar polling — detects meetings 30 minutes before they start
  • Per-attendee web research with up to 3 searches per person
  • 6 meeting type classifications: sales, vendor, partnership, internal, interview, customer success
  • 7-section Notion brief: Overview, Attendee Intel, Power Dynamics, Questions, Risk Flags, Outcome, Agenda Gaps
  • Cost scales with attendees: ~$0.12 (1) to ~$0.37 (3) per meeting
  • Built-in deduplication — same meeting never triggers twice
  • Full ITP test results with 20 fixtures and cost analysis
  • BQS v2 certification (12/12 PASS)

Frequently Asked Questions

How does the Google Calendar trigger work?+

The workflow polls your Google Calendar every 15 minutes. When it detects a meeting starting within the next 30–60 minutes, it fires automatically. No manual trigger needed — your brief is ready before you walk in.

What are the 6 meeting types?+

External Sales (prospect/customer calls), External Vendor (vendor evaluations), External Partnership (strategic partnerships, advisory boards), Internal (team syncs, 1:1s), Interview (hiring), and Customer Success (account reviews, onboarding). Intelligence adapts to each type.

How does per-attendee research work?+

The Researcher runs once for each meeting attendee using Sonnet 4.6 with web_search. It gathers seniority, company context, recent news, and LinkedIn signals — up to 3 web searches per person. Results are assembled before the Analyst synthesizes.

What does the Notion brief contain?+

Seven sections: Meeting Overview (type, attendee count, agenda summary), Attendee Intelligence (per-person research profiles), Power Dynamics (seniority map, decision-maker identification), Recommended Questions (3–5, meeting-type-aware), Risk Flags (job changes, company news, gaps), Suggested Outcome, and Agenda Gaps.

How much does each meeting cost?+

ITP-measured: ~$0.12 for 1 attendee, ~$0.27 for 2, ~$0.37 for 3. Cost scales linearly because the Researcher runs once per attendee. A team processing 20 meetings/month with an average of 2 attendees spends ~$5.30/month.

Why two different LLM models?+

Sonnet 4.6 handles per-attendee research where speed and web search matter. Opus 4.6 handles the synthesis where reasoning depth matters — classifying meeting type, identifying power dynamics, and generating context-aware questions. This balances cost and intelligence quality.

Does the same meeting trigger twice?+

No. Built-in deduplication uses a 7-day window via n8n static data. Once a meeting has been processed, it will not trigger again even if the polling interval catches it multiple times.

What calendar and delivery tools does this work with?+

Google Calendar for event source (OAuth2, calendar.readonly scope) and Notion for brief delivery (Integration token). It requires three credentials: Google Calendar OAuth2, Notion integration token, and Anthropic API key.

Related Blueprints