How I AIHow Webflow’s CPO built an AI chief of staff to manage her calendar and drive internal AI adoption
CHAPTERS
Why the “AI-native executive” matters (and why Rachel is unusual)
Claire frames the episode around a gap in AI discourse: lots of talk about AI-native ICs, less about AI-native executives. Rachel Wolan (Webflow CPO) represents a hands-on exec who uses AI tools daily and builds custom software to run her work.
Rachel’s on-ramp: from early coding to “vibe coding” as a daily driver
Rachel describes returning to hands-on building after years away from coding, enabled by modern app-generation tools. A personal weekend project (capturing kids’ memories) rekindled the habit, and she now spends significant time in Claude Code maintaining and iterating on her own app.
The AI Chief of Staff concept: building for an ‘N of one’ and iterating agents
Rachel introduces her AI chief of staff as the first app she tried to build—and has continuously refined. The goal is to offload repetitive executive workflows and to lead authentically by modeling daily AI prototyping behavior for her team.
Podcast prep workflow: generating a structured “How I AI” flow in minutes
Rachel demos how she uses Claude (often in multiple terminals) to generate a structured prep pack for speaking/podcast engagements. The output includes an executive summary and multiple narrative options, helping her be ready with minimal lead time.
Morning calendar triage: reflective analysis and planning delegation
Rachel shows how the chief of staff reviews her calendar to identify improvements and patterns—similar to what a human chief of staff might do. The assistant can flag issues like lack of customer time, poor energy management, and overbooking, then suggest concrete changes.
Under the hood & guardrails: API tokens, scoped permissions, and ‘janky’ power management
Rachel explains the technical setup: Google Cloud tokens stored in dotenv, with carefully limited permissions. She iterates on how much authority the agent has, often reducing power when it feels like overreach, emphasizing safety-by-design for personal agents.
Personalized UI design: note-card priorities, “Apple Notes” inspiration, and joy in software
They pause on the web app interface: a playful blue lined note-card UI for daily priorities and accomplishments. The chapter highlights “personal software” advantages—designing for delight, iterating quickly, and building ephemeral widgets that match how you think.
Brutal truth mode: using AI to coach executive behavior (and calibrating tone)
Rachel intentionally instructs the assistant to be harsh, producing blunt feedback like “you’re operating as a senior PM, not a CPO.” The segment shows how tone prompting can turn an assistant into a coach that surfaces uncomfortable but useful patterns.
Email triage agent: archive, prioritize, and draft—without replying to everything
Rachel walks through email workflows: the agent clears low-value noise, keeps critical items in the inbox, and drafts replies for high-leverage messages. The key challenge is teaching the agent what matters (e.g., partnerships) and what to ignore (e.g., SDR inbounds).
Networking dinner prep: screenshot → attendee research → tailored conversation starters
Rachel demos a dinner-prep workflow: she uploads a screenshot of the guest list, the agent identifies attendees, searches web/LinkedIn, and produces a structured prep doc. Because the system also has context about Rachel and Webflow’s latest releases, it can generate relevant talking points and priority connections.
Markdown as the knowledge backbone: personal context files and reusable agent memory
They describe storing key materials as Markdown inside the repo—personal communication preferences, resources, and monthly product release summaries. Markdown becomes both the app’s content source and a shared substrate that any agent (Claude Code/Cursor) can reference later.
Outcomes: closer to the metal, better technical conversations, and faster learning loops
Rachel reflects on how building the chief of staff improved her effectiveness beyond time savings. It deepened her understanding of AI product development, brought her closer to Webflow’s codebase, and made technical discovery more accessible without reading everything manually.
Driving internal AI adoption with “Builder Days”: structure, incentives, and sustained usage
The conversation shifts to organizational adoption. Rachel explains Webflow’s Builder Days—pausing normal work to build prototypes with tools like Cursor, Figma, Make, and Webflow—supported by warmups, office hours, judging panels, and prizes to push people past the initial friction.
Measuring impact and evolving talent systems: dashboards, feedback, ladders, and hiring
Rachel shares how they track adoption (e.g., Cursor usage dashboards) and collect survey feedback to iterate on the program. She ties AI fluency to broader talent systems—career ladders, interviewing, expectations—treating AI capability as a multi-dimensional competency, not a single skill.
Lightning round: product vs. personal software, and a practical prompting trick
In closing, Rachel predicts parts of an AI chief of staff could become a product ecosystem, though form factors may change. She shares a pragmatic prompting method: clear context when stuck, and use numeric intensity (“10x/100x”) to calibrate tone and output.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome