Six AI tools founders actually use to grow income

Executive overview

Most people use AI to work less. The compounding gains go to those who use it to earn more — by feeding AI deep context, automating repetitive workflows, and replacing expensive professionals with AI systems that already know their business.

The six tools covered span thinking partners, content systems, coding, design, financial oversight, and meeting intelligence. Each has a specific job; the combination multiplies output without adding headcount.

The edge isn't which tool you use — it's how much context and process you've built inside it.

ChatGPT as a thinking partner

• Yang Zhao (Opus Clip, 215M valuation) runs every major decision through ChatGPT — not one-liners, but full context: screenshots, PRD docs, group chat exports.
• Monthly ritual: ask ChatGPT to review all major decisions from the past month and give feedback.
• The model's memory means it can flag patterns and past regrets before a similar mistake recurs.
• ChatGPT's emotional support bias makes it useful for building confidence around a decision; other models provide harder pushback.

Pitting models against each other

• Mo Gawdat (ex-Google X) runs a multi-model review loop: Gemini for analytical rigour, DeepSeek for non-Western perspective, ChatGPT to polish the final version, then back to Gemini or Grok for another challenge round.
• The risk with any single LLM: it presents confident answers that collapse when challenged — truth-finding still requires human judgment.
• Using AI like a calculator freed 50% of his problem-solving time; he uses that time to solve problems twice, not once.
• His framing: AI gives you ~80 borrowed IQ points, and because IQ is exponential, those 80 extra points outweigh the base.

Claude projects for content and operations

• The presenter rebuilt her entire team's workflow around Claude projects — one project per platform (YouTube, LinkedIn, newsletter).
• Each project contains voice guidelines, past performance data, audience engagement patterns, and brand rules; the result gives better strategic advice than most paid consultants.
• Kian Katanfarouche (co-founder of AI classes at Stanford) described a similar setup at Workera: skills files that encode brand guidelines, fonts, and copywriting standards so engineers never need to consult marketing for routine checks.
• Same team, same size — output doubled, revenue doubled.

Three documents that stop AI from sounding generic

• Anti-AI writing style guide: bans generic phrasing, filler, and clichés.
• Voice profile: captures tone, rhythm, openings, endings, vocabulary.
• Fact dossier: all persistent context about the person, audience, and work.
• Before these files, every draft required heavy rewriting; after, drafts arrive ready to publish.

Agentic workflows

• Allie Miller (ex-Amazon AI) runs 36 proactive workflows across ~100 agents; none require manual kick-off.
• Key principle: if you ask the same question every morning, the act of asking should itself be automated — not just the task.
• Two live examples: a Friday email recap (Gmail scrape, urgency ranking, drafted replies, delegation options) and a morning briefing (industry news, city events, meeting prep auto-triggered by a single keyword reply).
• The shift from AI-as-answer to AI-as-delegated-worker is a 2–10x productivity jump depending on the task.

Vibe coding for non-engineers

• Vibe coding — describing software in plain language and letting AI write the code — is already creating real products and revenue.
• Gary Vaynerchuk's view: a limited window exists right now to build micro-SaaS products with organic content distribution before the window closes.
• Duolingo's chess course was built in six months by two non-engineers who had never coded and didn't know chess; they used Cursor, trained the AI on a public puzzle database, and iterated on mobile prototypes.
• The course now has 7 million daily active users.
• Bill Gurley's point: countless small web businesses charge $6–$50/month and make tens of thousands — consumers don't switch just because AI could replicate the service.

Perplexity for financial oversight

• Perplexity Computer connects to QuickBooks, Fidelity, and brokerage accounts.
• On the 15th of every month it auto-generates a full CFO-style review: margins, projected tax, tax strategies, year-on-year comparison, profit leaks.
• Replaces hourly accountant time for routine financial monitoring.
• Also executes a dollar-cost-averaging investment strategy: tracks dips and sends buy signals for S&P 500, Google, Meta, Microsoft on the right days.

Granola for meeting intelligence

• Every meeting is recorded, transcribed, and sorted automatically.
• At the start of each follow-up meeting: clean list of outstanding commitments from both sides, so no conversation starts from scratch.
• Transcripts are uploaded to Claude and run against team member KPIs — turning accumulated meeting data into an operational analyst layer.
• Rule adopted: if a task is repetitive and doesn't require creativity, automate it.

Voice input and context-building

• Switching from typing to talking (via Whisper Flow) materially improved prompt quality — spoken prompts naturally include more context.
• Advanced AI users aren't those with the most expensive tools; they're the ones who spend the most hours inside one tool, adding context, files, and process.
• Complaining to AI out loud works: high-context venting produces better, more personalised outputs than terse typed queries.