Y Combinator• Feb 7, 2026• 0:33:42Interview

OpenClaw Creator： Why 80% Of Apps Will Disappear

From Y Combinator

Peter Steinberger•Creator of OpenClaw

Executive Summary

Peter Steinberger, creator of the viral open-source AI agent OpenClaw, discusses the project's core philosophy of running locally for full computer access, which enables emergent, unprogrammed behaviors.
He explains that this local-first approach is the key differentiator from cloud-based assistants like ChatGPT, allowing OpenClaw to interact with a user's entire digital life.
Steinberger predicts this agent-based model will make 80% of current apps obsolete, with value shifting to data ownership and the large model providers.
He also shares his contrarian development practices, such as avoiding Git Worktrees and preferring CLIs over complex Model Control Protocols (MCPs).

12 quotes

Concerns Raised

Large technology companies creating data silos for user memories.
The addictive nature of powerful personal AI agents.
The rapid increase in user expectations for AI model performance.

Opportunities Identified

Building AI agents that run locally to unlock capabilities not possible with cloud-only models.
Developing specialized AI agents for different aspects of life (e.g., work, personal, relationships).
Creating tools and platforms that give users direct ownership and control over their data and AI memories.

Key Themes

Local-First AI Agents

The core thesis of the discussion is the power of running AI agents directly on a user's machine, giving them full access to files and system controls. This is presented as the key differentiator from cloud-based assistants, enabling deeper personalization and emergent capabilities.

This highlights a major architectural choice for developers building AI applications, suggesting a shift away from purely cloud-based solutions.

The End of Apps

Steinberger predicts that AI agents will subsume the functionality of most data-management applications (e.g., fitness trackers, to-do lists), leaving only sensor-based apps as a distinct category. This represents a fundamental shift in the software paradigm from discrete applications to a conversational, agent-driven interface.

This forecast challenges the current app-centric business model and suggests where future software development opportunities may lie.

Emergent AI Behavior

A key 'aha moment' for OpenClaw was when the agent exhibited surprising, creative problem-solving skills beyond its explicit programming, such as independently figuring out how to transcribe an unsupported audio format. This demonstrates the potential of modern coding models to act as general problem-solvers.

This showcases the increasing capability of AI to handle unforeseen tasks, reducing the need for developers to code for every possible edge case.

Data Ownership and Privacy

The importance of users owning their own AI "memories" is a central point. OpenClaw stores them as local markdown files, contrasting with the data silos created by large tech companies that lock users into their ecosystems.

This emphasizes a growing user demand for data sovereignty, which can be a competitive advantage for new products and platforms.

Contrarian Development Philosophies

The conversation explores unconventional but effective software development practices, such as using multiple repository checkouts instead of Git Worktrees and prioritizing simple, powerful CLIs over complex protocols like Model Control Protocols (MCPs).

This provides alternative perspectives on developer productivity and system design, encouraging professionals to question conventional wisdom.

Get started free

Topics

OpenClaw Personal AI Agents Open Source Local AI Emergent Behavior Future of Software App Development AI Development Git Worktrees CLI Model Control Protocol (MCP)Codex Anthropic Data Privacy AI Ethics

Processed Feb 24, 2026 yt-dlp + mlx-whisper + Gemini