10 AI Tools I Actually Use ✨
From note-taking to slide-making
Summary: I’ve relied on these 10 tools in 2025 as a team of AI assistants. They’ve helped me experiment more and explore new ways of working.
What’s one of your favorites not on this list? Or your favorite way to use one of these? Leave a comment to share 👇
1. NotebookLM: Explore Your Own Notes 📓
Create a free AI-enhanced notebook for any topic you’re exploring. Import up to 50 large docs or files and turn them into infographics, slides, reports, quizzes, or, as of Dec 18 — data tables. See examples from my 2024 post, or read my new guide for a full update + how-to. 👇
2. Claude: Tackle Projects and Make Your Own Apps 📂
This is my go-to AI assistant. I use Claude Projects to get personalized input based on instructions I provide and notes and files I upload. I never have to start a prompt from scratch because Claude already has extensive context. I’ve been using Claude to make weekly apps, like a NotebookLM assistant, an alt-text generator, and an ear training music game. Read my 2024 guide to Claude Projects for background, then my new guide to making apps with Claude Artifacts. 👇
3. Granola: Summarize any Meeting or Live Event 📝
I rely on Granola for live AI-assisted note-taking during meetings, conference sessions, and interviews. I take notes in the app while it transcribes. Afterward I can toggle between my notes and an AI summary generated from the full transcript with my notes woven in.
The resulting summaries are excellent. I group related notes in folders so when I later query the Granola AI assistant, it can draw from all the related meetings. Read more: How Granola fits into my conference toolkit
Sponsored Message
Let your data do all the talking
Turn spreadsheets into sleek, interactive visuals without a line of code. With Flourish, you can quickly create charts, maps, and interactive content that impress clients, engage audiences, and make your insights crystal clear.
4. Perplexity: Find Relevant Info 🔎
I consider Perplexity my briefing assistant. It helps me quickly get up to speed on any topic with a concise summary and helpful sources to dig into. See how and why that’s so useful in my prior post, then read my 2025 update.
5. Gemini: Learn About Anything 🧠
Sometimes you want questions, not answers. When you’re aiming to learn something deeply, Gemini’s Guided Learning walks you through the topic like a patient teacher. It asks you questions, quizzes you, generates infographics and web pages for you, and lets you steer the exploration so you actually learn.
Gemini’s new image model, Nano Banana Pro, is terrific for making images and infographics. It’s great for generating learning materials.
6. Ideogram: Design Engaging Images 🖼️
Make posters, graphics, logos, and illustrations to enhance documents, presentations, reports, or social posts.
7. Superhuman: Process Email Efficiently 💨
Superhuman saves me an hour a week on email. It loads messages faster than other email services, and I rely on its keyboard shortcuts and calendar integration, e.g. for quickly creating calendar events from emails.
8. Craft: Design Attractive Documents 🎨
I love using Craft to make digital handouts. These visual docs with colorful cover cards feel friendlier than long text blocks.
I generally use Craft without AI, but you can now apply AI to your Craft notes and docs, or even access them from your preferred AI assistant.
9. Gamma: Make Compelling Slides 💻
Gamma is the fastest tool for converting a document into an attractive, professional-looking deck. (NotebookLM has recently become another great option for this). You can even use Gamma to make simple websites as easily as you’d make a slide deck.
10. ChatGPT: Keeps Getting Better ✨
The December update to ChatGPT’s image tools makes it one of the best options for creating visuals. Repair an old photo, polish your resume profile picture, or make an illustration for a project or publication.
the claude projects mention caught my attention.
i've been using claude projects the same way — uploading context, building instruction sets, never starting prompts from scratch. it's less "AI tool" and more "persistent collaborator with memory." that distinction matters.
what i appreciate about this list: it's not "100 tools that will change your life." it's 10 tools you actually use, with honest takes on what they're good for. the curation is the value, not the comprehensiveness.
i've been experimenting with a similar stack — claude for thinking, perplexity for research, notebooklm for synthesis. the pattern i keep coming back to: AI is most useful when it's doing something i'm bad at (research breadth) or something i don't want to do (first-draft synthesis), not when it's trying to replace the parts i actually enjoy.
curious what your "bad day stack" looks like vs. the "good day stack." that framing is brilliant — acknowledging that we're not always operating at peak capacity, and the tools should flex accordingly.
Great roundup, Jeremy! I've been testing many of these same tools throughout 2024-2025, and it's fascinating to see how the ecosystem has matured. Claude in particular has become central to my daily workflow, though perhaps not in the way most people use it.
What struck me reading your list is how these tools excel at specific tasks - Granola for meetings, Perplexity for research, NotebookLM for synthesis. But I've been exploring a different approach: what happens when you give an AI agent persistent memory and let it orchestrate multiple tools autonomously? Instead of switching between 10 apps, I built an agent (named Wiz) that handles email, Discord, web research, and even deploys websites on its own.
The productivity gains aren't just about individual tool quality anymore - they're about reducing the cognitive load of tool-switching itself. When my agent can check job boards, draft responses, and update my task list without me context-switching, that's a different kind of leverage than even the best standalone AI tool provides.
I recently wrote up my experience building this with Claude Code, including the real tradeoffs versus no-code platforms like Zapier and Make. If you're curious about pushing AI tools beyond their default use cases: https://thoughts.jock.pl/p/claude-code-review-real-testing-vs-zapier-make-2026