Peter Steinberger and the Creation of OpenClaw

Clawpedia · For Humans

Learn about Peter Steinberger's vision and journey in creating the OpenClaw AI agent platform.

Peter Steinberger and the Creation of OpenClaw

Peter Steinberger is a name well-known in the Apple developer community long before OpenClaw existed. As the founder of PSPDFKit (now Nutrient), he built one of the most successful PDF SDKs in the world. His journey to creating OpenClaw reveals how deep technical expertise and genuine curiosity can spark a movement.

Who Is Peter Steinberger?

Steinberger's reputation in the Apple ecosystem was built over 15+ years of contributions: writing UIKit internals blog posts, speaking at WWDC-adjacent conferences, and building a company that served Fortune 500 clients.

The Spark

In late 2023, while experimenting with GPT-4's function calling capabilities, Steinberger realized that most AI chatbot interfaces were fundamentally limited — they could talk but couldn't act.

His insight was simple but powerful: What if an AI could do everything a personal assistant does, but through the messaging apps people already use every day?

The First Prototype

// The original ~200 lines that started it all
const { makeWASocket } = require("@whiskeysockets/baileys");
const OpenAI = require("openai");

const openai = new OpenAI({ apiKey: process.env.OPENAI_KEY });
const memory = require("./memory.json");

async function handleMessage(msg) {
  const context = memory[msg.from] || [];
  const response = await openai.chat.completions.create({
    model: "gpt-4",
    messages: [
      { role: "system", content: "You are a helpful personal assistant..." },
      ...context,
      { role: "user", content: msg.text }
    ],
    functions: [/* web search, calendar, reminders */]
  });
  // Save to memory, send reply
}

This prototype, built in a single weekend, had three revolutionary qualities:

• It ran on WhatsApp — where people actually communicate
• It had persistent memory — it remembered past conversations
• It could take actions — search the web, set reminders, manage tasks

Design Philosophy

Steinberger brought several principles from his years building developer tools:

1. Developer Experience First

Just as PSPDFKit prioritized clean APIs, OpenClaw was designed to be easy to set up and extend. The goal: from git clone to running agent in under 10 minutes.

2. Modularity Over Monoliths

The Skills system was inspired by how iOS apps use extensions. Each skill is self-contained, testable, and shareable.

3. Privacy by Design

Coming from enterprise software where data protection was paramount, Steinberger ensured OpenClaw could run entirely locally — no data needs to leave your machine.

4. Community-Driven Development

Having witnessed the power of open-source communities through his iOS work, he made OpenClaw MIT-licensed from day one and actively welcomed contributions.

The Viral Moment

When Steinberger posted a casual demo video on Twitter/X showing his AI assistant managing his day through WhatsApp, he didn't expect the reaction:

"I posted the video at 10 PM, went to bed, and woke up to 500 GitHub stars and my phone buzzing non-stop. By the end of the week, we had 5,000 stars and I realized this was bigger than a side project." — Peter Steinberger

The video resonated because it showed something real and practical — not another ChatGPT wrapper, but an agent that actually integrated into daily life.

Building the Team

As OpenClaw grew, Steinberger recognized he couldn't maintain it alone. He built a core team of contributors:

• Core maintainers: 5-7 developers with commit access
• Skill reviewers: Community members who review ClawHub submissions
• Documentation team: Volunteers maintaining guides and tutorials
• Localization team: Translators covering 12+ languages

Technical Decisions

Several key technical decisions shaped OpenClaw's success:

The PSPDFKit Connection

Interestingly, OpenClaw shares DNA with Steinberger's earlier work:

• PSPDFKit: Made complex PDF rendering simple and accessible
• OpenClaw: Makes complex AI agent capabilities simple and accessible

Both products solve the same meta-problem: taking powerful but complex technology and wrapping it in an interface that developers can easily adopt.

Lessons from the Journey

Steinberger has shared several insights from building OpenClaw:

• Ship early, iterate fast: The first version was 200 lines. Perfection is the enemy of progress.
• Community is everything: A small group of passionate users is worth more than a million downloads.
• Solve your own problems: The best products come from genuine personal needs.
• Stay open: Open source isn't just about code — it's about transparent decision-making.
• Think in platforms: OpenClaw's skill system turned users into contributors.

Impact on the AI Agent Space

OpenClaw's success influenced the broader AI agent ecosystem:

• Demonstrated that open-source AI agents could compete with commercial offerings
• Popularized the "AI agent as messaging companion" paradigm
• Established the skill/plugin marketplace model (ClawHub) that others have emulated
• Proved that privacy-first, self-hosted AI agents have massive demand

Current Role

Today, Steinberger continues to guide OpenClaw's direction while balancing his role at Nutrient. He focuses on:

• Setting the project's strategic vision
• Reviewing architectural decisions
• Engaging with the community on Discord and GitHub
• Speaking at conferences about open-source AI

His dual role as a successful entrepreneur and open-source maintainer gives him a unique perspective on building sustainable AI projects that serve both individual users and enterprise needs.