Webflow CPO Rachel Wolan’s AI Chief of Staff & Org Adoption Playbook

By: Rachel Wolan Host: Claire Vo Source: How I AI (ChatPRD) Type: podcast

Summary

Rachel Wolan, Chief Product Officer at Webflow, demonstrates three workflows that embody the AI-native executive approach. First, she built a custom “AI Chief of Staff” application running on her local machine via Claude Code, connected to Google Calendar and Gmail APIs with intentionally limited permissions (read-only calendar, read/archive/label/draft-only email). Each morning she prompts it to review her day and suggest delegations — it identifies meetings to make async, tasks to delegate to directors, and optional invites to decline. A unique “Brutal Truth” module provides unvarnished feedback on how she’s spending her time. Second, she uses a multi-step research agent for networking event prep: drop in a screenshot of a guest list, the system uses vision/OCR to extract names, then runs web search + LinkedIn search + targeted news search for each person, combining results with her personal markdown knowledge base (about_me.md, webflow_products.md) to generate a comprehensive prep document with priority connections, personalized conversation starters, and hot topics. Third, she shares her playbook for company-wide “Builder Days” — structured events with curated tool access, function-specific tracks, dedicated Slack support, warm-up exercises, judging panels, and prizes. After the first Builder Day, Cursor usage among designers went from near-zero to approximately 50% sustained adoption.

Key Ideas Extracted

“Brutal Truth” module for executive accountability: Programming an AI agent to deliver unvarnished feedback about time allocation provides the kind of honest coaching even human chiefs of staff hesitate to give
Permission guardrails by design: Read-only calendar, draft-only email (never send) — intentionally limiting agent permissions prevents mistakes while maintaining utility
Markdown-powered personal knowledge base: Simple .md files (about_me.md, webflow_products.md) provide personalization context without needing a complex database — low-tech, high-value approach
Vision/OCR for low-fi input: Dropping a screenshot of a guest list into an agent that extracts names via OCR eliminates manual data entry for networking prep
Multi-step research chains: Web search + LinkedIn + targeted news search per person mimics what a human would do but completes it for a dozen people in minutes
Builder Day structure for adoption: Curated tool access, function-specific tracks, dedicated support channels, warm-up exercises, judging panels, and prizes create a motivating environment that overcomes adoption friction
Measured adoption outcomes: Tracking tool usage before and after Builder Day with a Hex dashboard showed Cursor design adoption went from near-zero to ~50% sustained — moving the team from “laggard” to “early majority”
Personal experience builds leadership credibility: Building your own tools gives you the authenticity to lead organizational change — “improve yourself, then empower your team”

Notes

This is one of the strongest episodes for executive-level AI adoption patterns. The AI Chief of Staff concept (local app, API connections, intentional permission constraints, “brutal truth” module) is a replicable architecture. The Builder Day playbook pairs well with Brian Greenbaum’s Pendo episode as complementary approaches to organizational AI adoption — Greenbaum’s is more grassroots/ongoing, Wolan’s is more event-driven/top-down.

Raw Content

Rachel Wolan, CPO at Webflow, embodies the AI-native executive. She started coding at 16 and has returned to her builder roots. She walked through the custom “AI Chief of Staff” application she built — a personalized tool that helps manage the relentless pace of an executive schedule. Plus her detailed playbook for running company-wide “Builder Days.”

Workflow 1: Building a Personal AI Chief of Staff for Daily Triage

Problem: The sheer volume of information and commitments facing an executive. Prepping for meetings just minutes beforehand (“just-in-time executive”).

Step 1: Connecting the Dots with APIs

Google Calendar API and Gmail API connections
API tokens stored securely in .env file
Intentionally limited permissions: Calendar is read-only; Gmail limited to reading, archiving, labeling, and creating drafts (not sending emails directly)

Step 2: The Morning Calendar Review & Delegation Prompt Simple terminal prompt via Claude Code: “Tell me about my day tomorrow. What can I delegate?”

Agent returns prioritized suggestions:

Make Meetings Async: Identifies meetings that could be handled via document or Slack update
Delegate Tasks: Suggests specific meetings for directors to cover, even drafting the message
Decline Optional Invites: Flags meetings from unknown organizers or with unclear agendas

Step 3: Getting the “Brutal Truth” A unique programmed module for direct, accountable feedback. After analyzing the calendar, the agent delivered: “You’re operating as a senior PM, not a CPO. You’re reviewing PRDs, approving scripts, and recording marketing videos… The only thing that will matter in six months is [a specific strategic conversation].”

Step 4: Email Triage Managing a 500+ deep inbox:

Archive ruthlessly: Newsletters and non-essential communications
Highlight what’s important: Emails where someone is waiting on her for a document or decision
Draft responses: Prepared replies for recurring requests or partnership inquiries

Workflow 2: On-Demand Prep for Meetings and Networking Events

Step 1: Ingesting a Guest List with Vision Drop a screenshot of a dinner guest list into the app. The model uses vision capabilities to perform OCR, extracting names of every attendee.

Step 2: Multi-Step Research Agent For each extracted name:

Initial web search for broad overview
LinkedIn search for professional profile, current role, career history
Deeper targeted web search for recent news, articles, or talks

Step 3: Markdown-Powered Knowledge Base Agent is fed personal markdown files for personalization:

about_me.md: Professional bio, communication style, career highlights
webflow_products.md: Auto-generated monthly from release notes, detailing latest products and features

Step 4: Final Prep Document Comprehensive output including:

Priority Connections: Most relevant people to speak with
Personalized Conversation Starters: Tailored opening lines based on their background and hers
Hot Discussion Topics: Relevant industry trends (e.g., Answer Engine Optimization)
Venue Information: Restaurant details for easy reference

Workflow 3: The Playbook for Driving Org-Wide AI Adoption (Builder Days)

Step 1: Setting the Stage Goal: Get people over the initial friction hump.

Tool Access: Curated set — Cursor for coding, Figma for design, Make for automation, Webflow
Dedicated Tracks: Different assignments for various functions (Product, Design, Data Science, User Research)
Support Channels: Dedicated Slack channel staffed with engineers for troubleshooting

Step 2: Making it Fun and Engaging

Warm-up Assignment: Simple exercise to get comfortable before the main project
Judging Panel: Rachel and the CEO reviewed projects (friendly competition)
Prizes & Recognition: Winners awarded in different categories

Step 3: Measuring the Impact Used a Hex dashboard to track tool adoption before and after. Results: Cursor usage among the design team went from near-zero to ~50% adoption, sustained in following weeks. Moved a significant portion of the team from “laggard” to “early majority” on the adoption curve.

Post-event survey: participants found the day “fun, empowering, motivating, and eye-opening.”

Key Takeaway

The blueprint for the modern AI-native executive: start with a personal commitment to building tools that solve your own problems, then use that hands-on experience as authenticity and credibility to lead organizational adoption. Improve yourself, then empower your team.