“Every day, my Claude gets a little better”: how to automate your life with Claude Code
TL;DR
Hillary Gidley’s core automation test is brutally simple: if being 10x better at a task wouldn’t create 10x impact, automate it — she uses this for both work and life admin, from PowerPoint formatting to pediatrician scheduling, while protecting high-leverage work like writing, insight synthesis, and narrative-building.
Her ‘anti-system system’ avoids upfront setup and lets Claude Code learn from behavior instead of ideals — rather than prewriting preferences, she lets Claude observe real constraints like pumping, childcare coverage, and late-night work patterns, then update markdown files and calendars based on what actually happens.
The whole workflow starts with low-friction capture, not fancy tooling — Gidley uses an iPhone back-tap shortcut to dictate reminders like “reschedule pediatrician appointment,” which then land in an inbox Claude later organizes into a simple markdown-based task system.
She treats Claude like an employee: trust it to manage the back office, inspect only when needed — her daily planning flow pulls reminders, preferences, and calendar data, then breaks overwhelming projects like a baby passport into tiny next steps such as “just book the post office appointment.”
Her favorite integration is the ‘Yappers API’: narrate what you’re doing instead of wiring everything together — rather than overbuilding with OAuth and custom apps, she keeps Claude open beside her work and simply tells it what she’s seeing and doing, letting it maintain logs, reflections, and pattern detection.
One of the cleverest demos is a ‘recording on’ skill that anonymizes her real data live — instead of creating a separate demo environment, Claude swaps names and identifying details on screen while preserving the actual workflow, a trick with obvious value for B2B demos using production-like data.
The Breakdown
A new mom, a new business, and zero patience for wasted time
Hillary Gidley opens with the real constraint: her time has never been more valuable, and her attention has never been more fragmented. As a new mom and entrepreneur, she’s not chasing “hyper productivity” for its own sake — she wants AI to reclaim enough life admin so she can do meaningful work and still “go to the market and get a baguette.”
The ‘anti-system system’ starts with chaos, not perfection
She says a lot of AI productivity content loses her immediately because it assumes people want to build and maintain elaborate systems. Her answer is the opposite: start with the mess, admit you’re not a “Notion boy,” and let the system emerge from actual usage. The perfect visual for this is her favorite anecdote: a photo of Al Gore’s wildly cluttered desktop, which made her feel “seen.”
The first tool isn’t AI at all: a lock-screen capture habit
Before Claude enters the picture, Gidley solves the leakiness of memory with an iPhone shortcut: double-tap the back of the phone, dictate a task, and move on. “Reschedule pediatrician appointment” becomes a captured input instead of 2 a.m. anxiety, and she emphasizes how important it is that this step is incredibly easy to trigger.
Planning the day with Claude Code, markdown, and real-world constraints
In Claude Code, she types “plan my day,” and Claude pulls from reminders, a preferences file, and her calendar to build a usable schedule. The key point is that the preferences weren’t manually designed — Claude learned them by observation, including practical realities like pumping and weekend childcare, plus funny behavioral notes like “don’t do certain work before bed or you’ll get into flow and stay up too late.”
Why the passport finally gets done: Claude shrinks the task
One of the stickiest examples is the baby passport, a task she keeps meaning to do but avoids because it’s too big and fuzzy. Claude doesn’t schedule “get the passport done”; it schedules the first tiny step — just booking the post office appointment — which turns a looming burden into a 10-minute action that fits in the day.
The daily note and the ‘Yappers API’ close the loop
To track what actually happened, Claude creates a daily markdown note with the planned schedule and a running log of observed behavior. Gidley’s workaround for overengineering is what she jokingly calls the “Yappers API”: she keeps Claude open beside her work and just talks to it about what she’s doing, letting it notice patterns like “you list three priorities but only the first gets real time” and then adapt the system accordingly.
Complexity has to earn its keep
This is one of the video’s big philosophical through-lines: don’t wire up complicated integrations until the janky version proves itself. Gidley says she may only keep 20% of the workflows she experiments with, so building the fully connected version first is often a waste; screenshots, dictation, and narrated context are often enough to validate whether a workflow deserves richer automation later.
From demo privacy to returns: solve tiny problems with problem statements
In the lightning round, she shows a smart “recording on” skill that anonymizes live data during demos instead of forcing her to maintain a separate sandbox environment. She also walks through a returns workflow that started with a plain-English complaint — “I keep forgetting to return things on time” — and ended with Claude building a slash command, scripts, and retailer-specific logic, reinforcing her final message: you don’t start with a Python script, you start with a problem statement and use Claude every day until reaching for it becomes second nature.