The top 4 AI tools

The Hustle's June 4 issue featured a 90%-Claude, 10%-everything-else AI teacher, Grace Leung, breaking down when to use ChatGPT, Claude, Gemini, and Perplexity — and the prompts that actually work in each.

What You Need to Know: YouTube AI creator Grace Leung (100K+ subs) spends ~90% of her time in Claude and routes the rest to ChatGPT, Gemini, and Perplexity based on a three-question framework. Her toolkit includes 25 production-tested prompts and a clear map of which tool wins for content, data analysis, code, and research. The takeaway: there is no "best" tool, only the right tool per task shape, and the four leaders are diverging on capability rather than converging.

Why It Matters

• The "which AI tool should I use" question is now a $200/year-per-seat problem. Most pros are paying for at least three of the four. Picking the right default matters more than picking the best model.
• The four leaders are diverging, not converging. Claude for long-form reasoning and code, ChatGPT for breadth and ecosystem, Gemini for multimodal and Google-surface integration, Perplexity for grounded research. Picking one as your universal default leaves real money on the table.
• Prompt libraries are productizing. Leung's toolkit, Anthropic Academy, OpenAI's prompt guides, and Google's Workspace prompt gallery are all moving the same direction: opinionated, use-case-shaped prompts over generic "tips and tricks."
• A 90% default is a strong claim. If a working YouTube AI teacher with a six-figure consulting business lands on Claude for nine out of ten tasks, that's the strongest external signal yet that Anthropic's product wedge in professional work is real.

What Actually Happened

The 90% Claude user: Grace Leung's three-question framework

The Hustle's June 4 edition profiles Grace Leung, a digital growth consultant and YouTube creator with 100K+ subscribers, who built a six-figure consulting business around the four major AI tools. Leung reports using Claude approximately 90% of the time as her default, with ChatGPT, Gemini, and Perplexity handling the other 10% on a task-shaped basis. The Hustle promotes her AI Toolkit — gated download — as a packaged version of her decision framework plus 25 ready-to-use prompts.

Leung's three-question framework (paraphrased from the Hustle hook and toolkit landing page) asks, in order: (1) is this task long-form, multi-step reasoning (Claude) or short-form, single-step generation (ChatGPT), (2) does the output need grounded, citable sources (Perplexity) or synthesis from what I already know (Claude/ChatGPT), and (3) am I working inside the Google surface (Docs, Gmail, Drive) — in which case Gemini wins on integration cost. The framework is the part you actually want; the prompts are the part HubSpot monetizes.

Source coverage: The Hustle — "The top 4 AI tools? 🤔" (June 4, 2026), The Hustle — "The AI Toolkit I Use Every Week: ChatGPT vs. Claude vs. Gemini vs. Perplexity".

The four tools are diverging on capability, not converging

The "should I just pick one" question has been answered: no, and the gap is widening, not narrowing. Leung's breakdown — corroborated by every other working professional using these tools seriously — lines up with what the vendors are actually shipping:

• Claude (Anthropic) is winning long-form reasoning, code, document analysis, agentic workflows, and "I need one careful answer" tasks. Leung's 90% default maps directly to Claude's strength in multi-step planning, where it consistently out-performs on sustained, complex tasks.
• ChatGPT (OpenAI) is the breadth and ecosystem play: image generation, voice, custom GPTs, the largest third-party plugin surface, the most-used consumer product, and the best generalist for short-form generation and brainstorming.
• Gemini (Google) wins on Google-surface integration, native multimodal across Workspace, and price/performance at the low end. If your data lives in Drive, Gmail, or Docs, Gemini is the path of least resistance.
• Perplexity is the research specialist — grounded, cited answers, deep research mode, and now an SDK (Search as Code) for agentic search. It's not a generalist; it's a research instrument.

Source coverage: Perplexity research blog — "Rethinking Search as Code Generation", Google blog — "Introducing Gemma 4 12B".

The Take

If you remember nothing else: the right answer is a stack, not a default, and the people making money on these tools in 2026 are the ones who built a personal framework for routing between them, not the ones who picked the "best" one and stuck with it. Leung's 90% Claude number is striking because it's a default that gets overridden on real tasks; the most expensive mistake is treating your default as your only tool.

For the prompt-library debate: yes, they work, but the framework matters more than the prompts. A mediocre prompt with the right tool routing will out-perform a perfect prompt sent to the wrong model. Learn the framework, then borrow the prompts.

For vendor strategy: Anthropic's wedge in long-form reasoning and code is the most defensible moat in the consumer/prosumer AI space right now. If you're building on top of one of these, the bet is that "careful, multi-step, agentic" is the part of the workload that's growing fastest, and the part that the open-weights challengers (Gemma 4 12B, Nemotron 3 Ultra, Qwen3.7-Plus) are still catching up to.

Quick Summary

A working AI teacher (Grace Leung) routes 90% of her work to Claude and the rest to ChatGPT, Gemini, and Perplexity using a three-question framework. The four tools are diverging, not converging: long-form reasoning (Claude), breadth and ecosystem (ChatGPT), Google integration (Gemini), grounded research (Perplexity). Pick a default; don't pick a single tool.

Sources

• The Hustle — The top 4 AI tools? 🤔
• HubSpot / Grace Leung — The AI Toolkit I Use Every Week
• Anthropic — Claude overview
• OpenAI — ChatGPT product page
• Google — Gemini product page

Source: The Hustle (2026-06-04) | mr.technology — The Master Skill Index