I spend $2,000/mo on AI tools, here are the best ones

Key Concepts

Local AI Models, Voice AI, Context Engineering, Productivity AI, Automation AI, Web Search AI, Personal AI, Coding AI, Autonomous AI Agents, API Costs, IDE, UI/UX Improvement.

Local AI Models

• Olama: Easiest way to run local AI models (Deepseek, Gemma, Quen) via terminal. Command: Olama run Gemma 327B.
  • Benefits: Completely private, open source (fine-tunable), usable offline.
• Anything LLM: UI for interacting with local models, similar to ChatGPT.
• Apollo (Mobile App): Runs open-source AI models on phones (Deepseek, Lama 3.2, Quen 2.5, etc.).
• Cost: Olama & Anything LLM (Free), Apollo ($6 one-time).

Voice AI Tools

• Super Whisper: Voice AI tool with customizable vocabulary and keyboard shortcuts (Option + Tab).
  • Vocabulary feature allows adding custom words like "vectal".
  • Cost: $85/month.
• ChatGPT Voice (Mobile App): Translates foreign languages from videos (e.g., Arabic/French in Morocco).
  • Share video feature is OP for quick translations.
  • Requires ChatGPT Pro subscription ($200/month).

Context Engineering

• Repo Prompt: Simplifies context engineering by allowing users to load folders and craft prompts.
  • Loads files from a folder to provide context to the AI model.
  • AI-generates relevant files for context using models like Claude for Thinking.
  • Cost: $15/month.

Productivity AI

• Vectal: AI tool with custom system prompts optimized for specific projects (e.g., front-end web development).
  • One-click AI features like "idea cleanup" save time.
  • Users love the product.
  • Free to get started.

Automation AI

• NA10: Automation platform for creating workflows (e.g., tracking work ethic via Google Calendar and time tracking).
  • Automates repetitive tasks like checking email or collecting invoices.
  • Hostinger is used for hosting NA10 ($5/month).
• Lindy: Simpler automation tool with pre-built templates.
  • Cost: $50/month.

Web Search AI

• Perplexity: Fast and accurate web search tool that checks 10-20 sources quickly.
  • "Deep Research" feature checks 50-100 sources.
  • Faster than ChatGPT's default search and Google.
  • Cost: Pro plan ($20/month).

Personal AI

• GPT-4.5: Used for personal questions and life advice due to its deep world knowledge and human-like responses.
  • Estimated to have 12 trillion parameters.
  • Preferred by "high taste testers" (Sam Altman).
  • Cost: $75/million input tokens, $150/million output tokens.
• Claude.AI: Used for writing emails and copywriting due to its less recognizable AI writing style.
  • Claude models (Opus 4, Sonnet 4) are effective for generating human-sounding text.
  • Biased on controversial topics.
  • Cost: $100/month (5x limits), $200/month (20x limits).

Coding AI

• Cursor: IDE built on VS Code with AI features like background agents and bug bot.
  • Cursor Agent allows chatting with the AI and queuing messages.
  • Background agents support various models (03 Pro, Claude for Opus, Gemini).
  • Bug bot automatically runs on pull requests to spot bugs.
  • Cost: $200/month (Ultra plan).
• Codex (via ChatGPT Teams): Advanced autonomous cloud-based AI agent by OpenAI.
  • Generates four different versions of code implementations.
  • Highest pull request acceptance rate.
  • Used by the entire team for small tasks and bug fixes.
  • Requires ChatGPT Teams subscription ($270/month for 9 people) + ChatGPT Pro ($200/month).
• API Costs (OpenRouter, Anthropic API, OpenAI API): Significant expense, especially for using Claude GitHub action for pull request reviews.
  • Claude GitHub action costs $100-$150/month via Anthropic API.

Cloud Code

• Cloud Code: Best AI agent for coding productivity.
  • Workflow: Cloud Code on the left (implementation), Cloud Code on the right (consulting).
  • Used for quick UI updates and code tweaks.
  • Command: cloth update to get the latest version.
  • Example: Resetting a branch to match dev using get pull and force push.
  • Can 10x programming speed.
  • Cost: $600/month (3 subscriptions) + $100-$300/month (API usage).
  • Tip: Use Sonnet model to save money.

Synthesis/Conclusion

The speaker spends a significant amount ($2,000/month) on various AI tools to enhance productivity across different areas like coding, automation, and personal tasks. The most expensive and impactful tools are related to coding, particularly Cloud Code and Codex, which significantly boost development speed and efficiency. While some tools are free or relatively inexpensive (Olama, NA10 hosting), others require substantial investments (GPT-4.5, Claude, Cursor, API costs). The speaker emphasizes the importance of leveraging AI for both personal and professional gains, highlighting specific use cases and workflows for each tool. He also offers practical advice on cost optimization and model selection. The key takeaway is that strategic investment in AI tools can lead to significant productivity gains, especially in software development.