How OpenClaw's Creator Uses AI to Run His Life

By Peter Steinberger

~ 5 min read

player.exe -v "peter-steinberger-automation"

01_executive_overview.md

In this highly watched technical review, respected iOS engineering veteran and entrepreneur Peter Steinberger completely dismantles the notion that agentic AI is purely for "coding tasks." Instead, he demonstrates how it serves as a hyper-competent, tireless personal assistant for his deeply complex business workflows.

Peter Steinberger is not an AI researcher; he is a pragmatist. Having built and sold massive software companies, his time is his most valuable asset. When he evaluates a new tool like OpenClaw, he isn't looking for neat party tricks or theoretical benchmarks against GPT-4. He is looking for immediate, ruthless efficiency gains in his daily life. His video takes the audience far beyond the standard "hello world" scripts and plunges directly into the chaos of managing a high-stakes digital existence.

The core thesis of Peter's review is that the true value of OpenClaw lies not in its language generation capabilities, but in its ambient persistence. Unlike web-based Chat UIs where conversations die the moment a tab is closed, OpenClaw runs as a background daemon on his macOS machine, constantly vigilant, retaining state, and holding the keys (literally and figuratively) to his local file system, his email inbox, and his calendar.

02_the_email_triage_engine.md

Conquering the Inbox: Autonomous Triage

The most captivating segment of the video is when Peter demonstrates his morning routine. Instead of opening Mac Mail or Gmail, he asks his locally running agent to perform an action he calls "The Morning Sweep." For anyone drowning in hundreds of unread emails daily, the following sequence is nothing short of magical.

// Executing Workflow: The Morning Sweep

Peter:

"Run the Morning Sweep protocol on my inbox. Summarize the urgent issues, automatically archive newsletters, and draft polite declines to any cold-pitch sales emails."

OpenClaw Engine:

[Action] Invoking read_imap_inbox ('last_24_hours')

[Observation] Retrieved 142 unread emails.

[Thought] Processing emails via semantic categorization...

[Thought] Identified 84 newsletters/promos. Identified 12 cold sales pitches. Identified 4 urgent team alerts.

[Action] Invoking move_email_to_folder (target='Archive', count=84)

[Action] Invoking draft_email_reply (template='polite_decline', count=12)

[Action] Invoking generate_summary_report (target='team_alerts')

OpenClaw Response:

"Sweep complete. 84 newsletters archived. 12 drafts prepared for sales pitches (awaiting your final send approval). You have 4 critical alerts: A server downtime notice from AWS, and a PR review request marked 'urgent' from Sarah. Would you like the links to the PRs?"

What makes this sequence profound is the trust boundary. Peter explicitly mentions that he would never grant a cloud-based AI company like OpenAI or Google persistent read/write access to his company's IMAP server. The liability of a data breach is too high. However, because OpenClaw runs entirely locally on his self-hosted Apple Silicon hardware, the raw text of his confidential emails never traverses the internet to a third-party server. The reasoning happens locally, maintaining absolute corporate secrecy.

03_calendar_and_logistics_automation.md

The Logistics Matrix: Calendar Management

Beyond email, Peter showcases OpenClaw's ability to handle the logistical nightmare of meeting scheduling across multiple time zones. He demonstrates a common scenario: receiving an email from a client in Tokyo requesting an hour-long chat "sometime next week."

Instead of mentally calculating the JST to CET time difference, cross-referencing his Apple Calendar, and typing out three available slots, Peter simply forwards the intent to the agent: "Find a time for me to meet with Kenji next week. Propose three slots."

The agent intelligently accesses the local `ical` database, identifies his working hours, performs the timezone math, checks for conflicts, and generates an email draft proposing Tuesday 9 AM, Wednesday 4 PM, or Thursday 11 AM (Tokyo Time). It even generates the Zoom links in advance. The agent acts not just as a reader of data, but as a synthesizer of context, turning a five-minute friction point into a five-second approval click.

04_the_verdict.md

Final Thoughts: The Democratization of the Executive Assistant

Peter concludes his review with a philosophical observation. Historically, having a dedicated human executive assistant—someone who could anticipate needs, filter noise, and handle logistical minutiae—was a privilege reserved for C-suite executives and the ultra-wealthy. It required paying a full-time salary.

By proving that OpenClaw can handle roughly 80% of those exact same tasks utilizing local compute power that you already own, the technology effectively democratizes the "Executive Assistant" tier of productivity. It grants freelance developers, small business owners, and indie creators the leverage previously only available to CEOs.

His final verdict is absolute: OpenClaw isn't just a toy for nerds to write terminal scripts. It is a fundamental operating system upgrade for human productivity. The video ends not with a plea to write code, but with a challenge to the viewer: "Stop doing the robot's work. Let the agent handle the noise, so you can focus on the signal."