My current AI toolkit

Last updated 26 July 2025

Keeping up with AI tools these days is like trying to pick the winners of multiple races simultaneously. No one has time to keep up with the Tour de France and the Grand Prix and the Boston Marathon and whatever dumb horse race is the most famous.

These days there are multiple AI tools for everything, which to choose?

Well, I can’t tell you. No one can. But this list (which I update once a week or so) outlines what AI tools I use and why. I’d love to know how you get on with these tools and whether you use different ones! Let’s crowdsource the best ones.

Also: Here is a curated list of AI content creators I use to stay up to date.

What I need AI for

• Summarising blogs and academic papers
• Research
• As a thought partner.
• Editing LinkedIn posts, blog posts and reports
• Coding and data analysis
• Productivity: speech to text, taking meeting notes
• Quickly answering my many random questions every day
• Making pictures, usually for presentations, blog posts and fun
• I do not really do any video generation, so this list won’t have that.

General use tools

• Unsurprisingly, ChatGPT is my main AI tool. I subscribe to the £20 plus tier, and recommend that you do too (or get Gemini advanced, both good choices). It has so many killer features, which I use daily:
  • o3 is the smartest model on the market
  • Deep Research: Worth the £20 on its own.
  • Projects: I make projects for everything, I have 10 at the moment! Every research project gets a ChatGPT project. Other include an image generator that follows my branding, to one that helps me with a roleplay game I run with my friends.
  • Custom GPTs: I have a custom GPT to help me with blog posts, one for data visualisations, another for reviewing my LinkedIn posts. Quite a few!
    • I wrote a post about to easy way to get value out of AI using projects and GPTs here.
  • ChatGPT Advanced voice mode: Video chat with ChatGPT. They improved it recently and its honestly more gratifying to talk to than most humans. I use it for brainstorming mostly. Other killer use cases involve having it tell you how to navigate terrible websites, live translate, or tell you what button to push on your mate’s washing machine because you can’t get it to work.
• That said, I also love Google Gemini 2.5. It’s fast and smart.
• For AI search engines I mostly use Perplexity on the free tier.
• As I only do casual image generation, I use ChatGPT.
  • For fast images I use Gemini, but its not great at following instructions. I mainly use it inside Google slides because its fast and does 4 images at once.
  • It has a cool feature where if you ask it to update an image that is already generated, it’s pretty good at doing that. If you make an image and the person has brown hair, you can say “make the person have blonde hair,” and it will remake that image rather than generating a whole new one
• For Deep Research, I use ChatGPT, Gemini and Perplexity.
  • ChatGPT thinks deeper, Gemini searches broader.
  • I have written more on Deep Research tools here.
• For coding in VSCode, I use Github Copilot, on the paid tier.
  • I have had some insane results with this tool, where it predicts the function I want to write before I write it. It’s far easier than copying and pasting code back and forth into another tool; you can ask it questions about your code and it will rewrite the code write in your IDE.
  • I use “agent mode” to code up “micro software” multiple times a week. Little scripts here and there to automate small, tedious tasks.
• I also use Google’s Jules, which is a free asyncronous developer that you attach to your github repo and have it write code for you while you sit in meetings and make coffee. It’s not amazing, but its free and you can always just ignore it’s work.
• For Text to speech on my phone I use the Speechify. Give it a URL, and it will read it to you fluently at any speed you like. Good for when I have eye strain from reading too much.
  • I used to use the Reader App by Eleven Labs, which is much better than Speechify, but they upped the price a lot.
• I use Snipd for AI-powered podcast listening. Snipd has an amazing feature where it will save clips of the podcast when I tap my earbud play button 3 times, allowing me to quickly save interesting snippets for later.
  • It uses AI to select a clip that captures when the speaker started making the point to when they finished, intelligently determining how long the clip should be.
  • You can also import YouTube videos into it if you want to just listen to a YouTube vid and “clip notes”.
• For transcribing my speech to text, I use SuperWhisper. I have tried many of these apps over the years with varying success, but the quality of this one (for the fact that it’s got a free plan) is staggering. It removes filler words, understands rare names (like my wife’s name “Mizla”), and even knows to abbreviate some terms. I once said “Auto Regressive Integrated Moving Averages model, and it intelligently transcribed it as ARIMA, which shocked and delighted me.
• To help me organise my research notes, I use Obsidian.md which isn’t an AI tool, but I do make heavy use on an AI plugin called Smart-Connections. When you open a note, it recommends other notes of yours that are conceptually similar. It helps you spark connections and identify duplicate notes.
• Google’s NotebookLM is another brilliant tool. It has so many great uses, but the key ones for me are:
  • When doing research projects, you can upload up to 50 documents to it and then chat with them.
  • Turning a collection of reference documents (i.e. SOPs, guidelines) into chatbots.
  • You can convert a document or documents into a podcast episode where 2 AI “hosts” discuss the content. This really wowed me with how natural it is.
• I use this Firefox plugin for summarising YouTube videos. I do this with longer videos to decide if they’re worth watching (there’s so much clickbait on YouTube). What I love about this extension is that it does it right in the YouTube window, and you can choose a short, medium or long summary.
• When it comes to finding academic papers, I use Elicit, and Scite AI.
  • However, I don’t get huge value out of these tools.
• I use Granola as my AI meeting software.

Great tools I used every so often

• I use GigaBrain as a search engine that specialises in Reddit. I love this because in my experience Reddit is 10% “expertise and depth of analysis you rarely see anywhere else” and 90% garbage. Skip the garbage, use GigaBrain.
• SciSpace and Consensus are excellent tools for finding academic papers
• Google Fact Check Tools: A super cool tool from Google that is basically a search engine but for Third party fact checks. If you search for something it will return articles about that thing, but also crucially what fact-checkers said about that article.

Tools I used in the past

• Wispr Flow It’s definitely one of the best text-to-speech programs, but because it sends your voice clips to the cloud, I don’t use it as much. My job requires me to use a model that is stored locally on my computer.
• Note GPT does a few things, but I only really used it to easily summarise long YouTube videos to figure out if they’re worth watching. It was good enough, but has a limited free tier, and now Google NotebookLM can do it just as good for as many videos as I’d like.
• Storm is a tool from Stanford that produceds first draft wikipedia pages on anything you like. It’s great when you’re first diving into a topic, but since I started paying for Gemini Advanced Research, it’s less necessary.
• Over the last year I paid for Perplexity Pro and You.com, premium tier, but ultimately did not use the premium features enough to justify it. Perplexity Pro is great, and I love You.com’s premium feature of letting you use other models, but I just don’t use it enough.
• I have at various points used the coding assistants Supermaven and Codeium. Both are great and free, and I haven’t noticed any differences in code quality between them and Github Copilot. But Copilot has a more convenient UI.

Cool tools I don’t use

• Claude Artefacts seems extremely cool but again, I don’t use it enough.
  • Some people reportedly use it a lot.
• Otter.ai and Fireflies.ai are both great AI meeting tools that seem as good as Granola. I just found Granola first, and haven’t got the time to test them against each other. If you have a clear preference, let me know!
• Gamma Makes PowerPoint presentations using AI. It will generate the images, it will generate the formatting. You just give it an outline of what you want to say and it will turn it into a beautiful presentation. I’ve used it a little bit but in the end I wrote my own Python script to do this for me, for free.

Tools I am intending to try at some point

Anyone who’s used these, please let me know if they’re any good

• Make.com is a great tool for making automations.
• Thesisify: Is an AI academic research and writing tool I’ve been recommended.

If you’d like a tailored workshop for your non-profit, where I guide you through getting the most out of AI, get in touch on LinkedIn or using my email above. I’ve helped numerous non-profits, and I’d love to help you.