OpenClaw Just Killed 80% of Apps. Here Are the 5 That Survive.

TL;DR

OpenClaw lets AI agents operate computers autonomously, threatening most apps — but 5 software categories remain viable because AI can't replicate data, trust, or human relationships.

Key Points

• 1.What OpenClaw is: An open-source AI agent that controls its own Mac Mini, browser, and apps, receiving instructions via WhatsApp/Telegram — it acts rather than just answers, and its creator Peter Steinberger just joined OpenAI.
• 2.The trigger event: Anthropic's Claude Co-work plugins caused a "SaaS apocalypse" — $285 billion wiped from public software stocks in 48 hours, approaching nearly $1 trillion in total market value lost.
• 3.What dies: Thin UI wrappers on databases, basic CRMs, single-feature apps, simple schedulers/invoicing tools, and lifestyle/habit-tracker apps — anything with no data moat, no community, and no agent integration.
• 4.Survivor #1 — Data Gold Mines: High-quality, structured, niche databases agents can query (e.g., building material price aggregators, local childcare availability trackers, restaurant food cost databases).
• 5.Survivor #2 — APIs: Every agent action is an API call; niche APIs (like booking availability for hair salons) become more valuable as agent usage scales, similar to how Stripe and Twilio grew.
• 6.Survivor #3 — Trust/Creator Apps: AI can copy features but not relationships; creators and coaches building apps tied to their specific methodology retain a low-acquisition-cost audience moat.
• 7.Survivor #4 — Network Effect Platforms: Marketplaces, niche communities, and professional forums where value lives in the people, not the software (e.g., vetted tradesperson lead-swap platforms, local parent review communities).
• 8.Survivor #5 — Vertical AI Agents: Deeply specialized agents for one industry (e.g., dental, brewery compliance) that accumulate proprietary data over time — plus a bonus category: agent-to-human labor marketplaces where AI agents hire humans for tasks.