One VC Spends $120 Monthly On AI. Beats Having An Intern. The Method Matters, Obviously.
Marc Palet, a 26-year-old tech VC and sole employee outside the US at his firm, pays over $120 monthly across Claude, ChatGPT, n8n, and transcription APIs. He has invested in over 12 AI companies and credits his current productivity stage to Andrej Karpathy's concept of building a knowledge base that automatically processes information. Despite extensive AI use, Palet notes it has not replaced his existing software subscriptions but added new ones.
This illustrates the principle of compound tooling. You do not replace your stack with AI. You layer AI on top of existing workflows to eliminate bottlenecks. The mental model here is orchestration over substitution. Palet is not firing software. He is conducting it. The Karpathy insight about automated knowledge bases is the real lesson. Stop thinking of AI as a chatbot and start treating it as a processing layer that sits between you and your information.
Marc Palet, a venture capitalist who has invested in over 12 AI companies, uses Claude, ChatGPT, n8n, and transcription APIs as the sole employee of his firm outside the US.
- List every software subscription you currently pay for and identify one repetitive task in each tool that eats your time. Expected outcome: a concrete inventory of automation opportunities.
- Sign up for n8n at n8n.io and connect it to two of your existing tools, such as Gmail and Google Sheets. Build a simple workflow that logs incoming emails into a spreadsheet automatically. Expected outcome: your first automated pipeline running in under 15 minutes.
- Open Claude or ChatGPT and ask it to draft a knowledge base structure for your work, organized by recurring task type. Feed it your most common documents and ask it to generate a tagging system. Expected outcome: a searchable knowledge architecture you can maintain over time, mirroring the Karpathy approach Palet adopted.