The AI Tools I Use for Work
Here are a few updated notes on the different AI tools I use to support my work as of November 12th, 2025.
This list is a snapshot of my subjective preferences that will likely be revisited.
I hope you find it helpful.
PS: The next cohort of my AI literacy for business course starts on November 19th in case you'd like structured guidance on implementing AI tools to support your work.
Chatbot
If you've heard me speak in the last two months, you should know that Claude is my go-to chatbot for productive AI collaboration.
I said it at Elevate. I said it at BFN. I said it at BFUTR. I will keep saying it until it is no longer true, which might be in another two weeks based on recent rumours.
Why? Because Claude gets a fraction of the web traffic ChatGPT and Gemini do, but more importantly, it tells me when I'm wrong or not thinking hard enough.
The result: I've rarely reached for ChatGPT or Gemini since Claude Sonnet 4.5 was announced on September 29th.
GPT-5.1 was released as I was typing. Its default tone irritates me, much like GPT-5 and GPT-4o did. I'll try the professional mode customization later this week.
Search Engine
I use Kagi as my search engine of choice. Think of it as a better Google search, with:
- No ads
- User-personalized results
- Filters for different parts of the internet
- Optional AI quick answers and webpage summaries
- A daily personalized news digest
- AI search assistants with custom prompts and AI model selection
I'm currently using a custom prompt with the GLM-4.6 (Reasoning) model for my AI mode, and I've noticed the level of detail in my prompts directly affects the depth and quality of the responses I get.
This is why I have a prompt I use to improve my initial prompts.
Speech to Text
I can talk much faster than I can type, and when working with AI, the more information I input, the better the outputs.
I'm currently using Wispr Flow (Affiliate Link) because the dictation is near instantaneous, and it automatically formats the text based on the app I'm currently using. It also works on my phone, helping me dictate texts and emails on the go.
Software Development
I use v0 for prototyping user interfaces or new features for Edventive and other projects of mine.
I use Cursor and Claude Code for serious development work.
Cursor has the quickest and most useful code autocomplete I've tried for React and TypeScript codebases. Their new model, Composer-1, impressed me with its speed. I use it for small, well-defined fixes or refactoring.
Claude Code has been the biggest surprise I've experienced this year. Think of it as using a chatbot connected directly to your codebase. I work with it to plan features, bug fixes, and code quality improvements, then set it off to work while I focus on more important tasks than typing.
I should note that I spent months creating relevant rules and examples for it to follow, on top of reorganizing my codebase to make it more modular and composable.
I haven't asked it to build anything I couldn't yet, because I need to understand the code, verify it works, and then understand what to change if it doesn't.
Image Generation
Midjourney is my preferred text-to-image tool for creative expression. It helped me with ideation for the Edventive and Upskailed logos and created the photos I use in my pitch decks.
I'm convinced I'm not using more than 2% of its capabilities.
I haven’t had much use for image editors like GPT and Nano-Banana lately, so I can’t comment on my preference.
Power User Tools
Raycast is my everything tool. It fundamentally changes the way I use my computer. Between app launching, text snippets, clipboard history, and AI commands, I can't go back.
Chorus is my deep work chatbot. I like to use it to chat with multiple large language models (including Ollama models) at the same time so I can compare their outputs and pick the best ones for each task I'm working on.
New Discoveries
Ollama is my latest and greatest discovery, and yes, I know I'm late to the party.
Ollama lets me download open-source AI models and use them locally on my computer without needing internet access or sending any of my chats to the cloud. Think of it as installing a large language model on your computer, similar to how we used to install old-school software from CDs.
When performance matters more than absolute privacy, Ollama Cloud allows me to chat with the best open-source models available, including Kimi K2-Thinking and GLM-4.6 Plus, both of which have impressed me lately.
Rumor Mill
I’m eagerly anticipating Gemini 3, Google's new family of models, that is expected to launch within the next few weeks.
Gemini was the best model in the world at several points this year, but has noticeably fallen behind Claude and ChatGPT's performance in Q3 of 2025.
Suggestions
At this point, the capabilities of current AI tools far surpass most people's ability to make the best use of them, including my own.
I strongly believe that most people and organizations would benefit more from learning to use these tools effectively than from switching to the latest and greatest tool every time a new one comes out.
Last Call: The next cohort of my AI literacy for business course starts on November 19th in case you'd like this kind of guidance.