How I AIHow AI got me 3 promotions: the ultimate guide for EAs (w/ Zapier’s EA)
CHAPTERS
Why EAs should embrace AI: from repetitive work to higher-leverage impact
Claire introduces Cortney Hickey (EA to Zapier’s CEO) and frames the episode as a hands-on guide to using AI for real admin and ops leverage. Cortney explains her motivation: eliminate repetitive work, stay ahead of the curve, and reshape the EA role into more strategic, creative, relationship-driven work.
- •AI as a way to “work yourself out” of boring/manual tasks
- •Mindset: adoption is a “when, not if” for EAs
- •EAs’ calendar-centric work makes them ideal early adopters
- •Goal: reclaim time for higher-impact and more human work
Building a weekly meeting-prep agent in Zapier Agents (what it automates)
Cortney walks through her flagship workflow: a scheduled weekly meeting prep agent that scans the upcoming calendar and prepares research and context. The agent replaces a multi-hour Friday routine of prep and information gathering across tools.
- •Scheduled weekly trigger (e.g., Friday 8am) for next-week prep
- •Filters meetings that truly require prep (vs routine 1:1s/standups)
- •Web research for external participants without Zapier emails
- •Connects to CRM (HubSpot) to pull relationship/deal context
The agent’s “second brain” workflow: calendar + CRM + email + Slack context
The meeting prep agent mimics what an EA would do manually: check attendee background, CRM history, and internal communications to assemble a single briefing. Cortney emphasizes this reduces scattered searching and prevents critical context from living only in her memory.
- •Looks up participants by email/company in CRM for sales notes and relationship status
- •Searches Gmail for prior threads and Slack for company mentions/context
- •Consolidates prep into one place instead of hopping across tools
- •Acts as memory augmentation: fast refresh on people and meeting purpose
Outputs that drive action: Todoist tasks + Slack weekly digest with insights
Cortney explains the two core outputs that make the automation actionable: prep tasks created in Todoist and a structured Slack digest. Beyond summarizing facts, the agent generates recommendations on what to prioritize for the week.
- •Creates a dedicated Todoist task per meeting with gathered intelligence
- •Schedules prep tasks relative to meeting time (e.g., 2 hours before)
- •Sends a Slack digest summarizing key meetings + any errors/unknowns
- •Generates “prep recommendations” (e.g., review past deck, check PR priorities)
Improving agent quality over time: traceability, feedback, and “progress over perfection”
Cortney shows how she inspects the agent’s step-by-step reasoning and iteratively upgrades it (e.g., adding LinkedIn links). The emphasis is on starting simple, then refining based on real usage and observing where the agent fails or misses context.
- •Review the agent run log to see what it did and why
- •Use Copilot/natural language to add features (e.g., LinkedIn hyperlinks)
- •Start basic; expand capabilities as needs become clear
- •Keep updating as tools and integrations improve
How to think about agents: narrate your process and train an “intern”
Claire and Cortney discuss why agents are approachable: you can simply describe the steps you would take. Cortney likens agents to interns—initially imperfect, but trainable to learn your systems and eventually operate reliably with minimal supervision.
- •Design by narration: “First I’d Google… then check Slack… then CRM…”
- •Agent as intern metaphor: teach your system and standards
- •Templates can be shared org-wide and customized per person/tooling
- •EAs can scale their expertise beyond 1:1 support
Beyond prep: calendar optimization and personal-professional constraints
Claire adds an extension to the workflow: use agents to find calendar issues and protect time (including personal constraints). The chapter highlights that meeting support isn’t just research—it can also improve schedule health and sustainability.
- •Identify back-to-back blocks, missing breaks, or no lunch windows
- •Recommend combining or removing duplicative meetings
- •Block protected time based on preferences (e.g., morning school drop-off)
- •Use the same system for “Sunday Scaries” planning and optimization
Culture reinforcement via automated meeting feedback (from Fathom experiments to org rollout)
Cortney shares how she began prompting meeting transcript tools to generate coaching feedback tied to team effectiveness frameworks. Over time, this became an automated system that normalizes feedback and reinforces Zapier’s operating principles.
- •Started manually prompting meeting notes for coaching-style feedback
- •Used frameworks like ‘Five Dysfunctions of a Team’ to guide evaluation
- •Feedback normalizes growth and reduces anxiety around receiving critique
- •Automation helps reinforce values continuously in a remote org
Designing the automated meeting coach: tone calibration, guardrails, and delivery
The team refines the coach to balance being “demanding and supportive,” including making feedback more direct when needed. Cortney outlines guardrails (meeting length, participant type, sufficient context) and automated delivery to each participant via Slack mapping.
- •Tune feedback tone: avoid overly soft coaching; make it actionable
- •Guardrails: skip short meetings; ensure enough context; employees only
- •Use company values, meeting norms, and impact behaviors as rubric
- •Match emails to Slack and send AI-generated feedback from a bot
Stress-testing strategy docs with an “exec prep” GPT (scaling Wade-style feedback)
To prevent herself becoming a bottleneck for strategic docs, Cortney builds a ChatGPT-based reviewer that critiques and strengthens decision documents before they reach execs. The GPT helps writers clarify purpose, surface trade-offs, and tighten recommendations—sometimes eliminating the need for a meeting.
- •Problem: many people want EA help anticipating how the CEO will react
- •Solution: GPT that reviews “TP” (strategy/decision) docs for exec readiness
- •Feedback includes clarity fixes, trade-offs, data gaps, and rewrite examples
- •Benefits: higher-quality async work, fewer meetings, less EA bottlenecking
What powers the exec-review GPT: internal knowledge, examples, and adoption proof
Cortney explains the knowledge base behind the GPT: strategy memos, norms, good examples, and “managing up” guidance. She highlights why building it in ChatGPT encourages psychological safety and shows adoption metrics as evidence of org-wide enablement.
- •Knowledge inputs: values, norms, roadmap, strategy memos, strong TP examples
- •Includes “Wade feedback tuning” and managing-up guidance for realism
- •Built where users feel comfortable sharing drafts without social friction
- •Usage analytics demonstrate scale and impact across the org
NotebookLM as a centralized “strategy companion” (interactive strategy + AI-generated podcast)
Cortney describes a company-wide NotebookLM workspace that aggregates dozens of strategic sources (docs, all-hands, transcripts, org action plans). Employees can query it for personalized guidance and even consume strategy via auto-generated audio summaries.
- •Centralizes strategy across many sources into a single queryable system
- •Helps employees connect big-picture strategy to their specific roles
- •Continuously add sources to keep answers current and comprehensive
- •Supports multimodal learning, including AI-generated podcast-style summaries
Lightning round: AI won’t replace EAs—it promotes them; limits of “AI exec clones”; prompting style
In closing, Cortney argues AI elevates EAs by removing time constraints and enabling deeper, broader organizational impact—citing three promotions as proof. She notes partial executive replication is possible (writing style, norms), but keeping up with a fast-evolving leader’s thinking is difficult; her prompting style is direct, detailed, and iterative.
- •EA role expands: more depth across the org, more time for strategic projects
- •Career impact: Cortney credits AI leverage with multiple promotions
- •Exec cloning works for stable patterns; harder for rapidly evolving judgment
- •Prompting technique: be direct, ask for reasoning, and give candid feedback
