Claude
Think, analyse and write
“My favorite AI for most tasks.”
Full reviewMy tool reviews
AI Agents
Autonomous AI frameworks and systems that run commands and manage workflows in the background.
All tool reviews5 reviews
No matching reviews
Try a shorter name or clear the category filter.
Tools and models
ChatGPT
An assistant for everyday tasks
“ChatGPT has earned its place back on my machine.”
Full review
Gemini
Get quick answers and visual feedback
“The biggest plus for me is the near-instant answers.”
Full review
Gemini Notebook (formerly NotebookLM)
Understand your own documents
“I keep coming back to it when I need to actually absorb something dense rather than just skim it.”
Full review
Wispr Flow
Speak instead of typing, and take meeting notes
“I barely type anymore.”
Full review
Suno
Make a personal song
“I've had a lot of fun making songs for my daughter and friends.”
Full review
Higgsfield
Try image and video models
“I like having different models and creative controls in one place”
Full review
Lovable
Turn an idea into an app
“It is also great for quick prototyping and for figuring out difficult UX/UI questions.”
Full review
Atlassian Rovo
Work with Jira and Confluence
“It doesn't feel really smart yet, but within a small, well-defined context, it's quite helpful.”
Full review
Claude Code
Build and maintain software
“I use it to build and maintain this entire website”
Full review
Claude Cowork
Delegate files and office tasks
“Give it a folder of messy notes, come back to a structured report.”
Full review
Claude for Excel
Work on spreadsheets with Claude
“your highly proficient assistant for any spreadsheet task.”
Full review
Claude for PowerPoint
Draft and edit PowerPoint slides
“it saves real time on the fiddly parts.”
Full review
Claude for Word
Edit and format Word documents
“Chat is now the primary interface I'd use for formatting my Word documents.”
Full review
Claude in Chrome
Let an agent work in your browser
“it has its charm, because it's so simple.”
Full review
Copilot
An assistant inside Microsoft 365
“The problem: it's not productized well.”
Full review
ElevenLabs
Turn writing into spoken audio
“Hand it a written dialogue and it'll voice it with multiple speakers, emotions, and audio effects.”
Full review
Figma Make
Explore designs and user flows
“It's fantastic for creating prototypes that show entire user flows.”
Full review
FLUX
Generate and edit images through an API
“I reach for FLUX when I need images inside a pipeline rather than clicking around in an app.”
Full review
Google Antigravity
Build apps and work on your computer
“I enjoy using it to experiment with websites.”
Full review
Google Jules
Run scheduled checks on a code repo
“it has a look at my web projects and sends me an email from time to time.”
Full review
GPT-Image-2.5
Create and edit images in ChatGPT
“now this is my daily driver.”
Full review
Grok Bot
A coding agent for technical users
“my go-to when I've hit my token limits with the other models.”
Full review
Grok Imagine
Make images and video inside Grok
“interestingly it has caught up with the frontier in many respects.”
Full review
Hermes Agent
Run an agent that builds reusable skills
“Hermes has a built-in learning loop, so it actually builds and saves reusable skills as it works.”
Full review
Midjourney
Explore concept art and visual styles
“It's likely more suitable for design professionals who can invest the time in prompt craft.”
Full review
n8n
Connect apps and automate workflows
“n8n is the backbone of our automation stack.”
Full review
Nano Banana
A second opinion on images
“it's quite good now with a lot of things.”
Full review
OpenAI Codex
Build software with an AI agent
“depending on your tech stack, Codex might be the better, faster, and currently especially the cheaper choice.”
Full review
OpenClaw
Reach your AI assistant through chat
“of the personal assistants I've tried, it's the only one I still use regularly.”
Full review
Perplexity
Research a question with sources
“if you research for a living or need sources you can really rely on, a dedicated research tool is still worth paying for.”
Full review
Perplexity Computer
Research a topic and build an app
“It's too expensive for what I get out of it.”
Full review
Poe.com
Use different AI models in one place
“Minus points for bad user experience and UI that hasn't been improved at all in two years.”
Full review
Replit Agent
Build and publish an app in a browser
“one of the most accessible ways to go from 'I have an idea' to 'here's a working app.'”
Full review
Replit Design
Try a clickable UX prototype
“I use Replit mostly for Replit Design.”
Full review
Stitch
Try another approach to a UX problem
“a useful place to get a second or third opinion on a UX question”
Full review
Superwhisper
Dictate privately on your own device
“Several of my colleagues use Superwhisper, and they still like it.”
Full review
UX Pilot
Turn requirements into UI screens
“I've lowered it to 4 because it hasn't kept up with other tools in my workflow.”
Full review
Veo 3.1
Animate images and create video clips
“I love it to animate images I've generated”
Full review
Zapier
Automate tasks without a server
“If you've never used workflow automation before, start here, not with n8n.”
Full reviewKeep learning
The courses and documentation I recommend when you want to go further.
Claude Academy & Documentation
Learn to get more from Claude
“I look at it frequently and recommend it to all my coworkers.”
Full review
Google AI Essentials
Get a structured introduction to AI
“The best 'send this to your colleague who hasn't started yet' resource I know.”
Full review
OpenAI Academy
Learn to use ChatGPT with a purpose
“Solid starting point if ChatGPT is your primary tool.”
Full reviewMy ratings are personal judgements. Open a quick look for the reasons.
Claude
Think, analyse and write
My take
My favorite AI for most tasks. I use the standard Claude for conversations, and I get the most out of it in agentic setups: Claude Code for autonomous coding, Claude Cowork for longer workflows.
What it's good at
- Best-in-class for nuanced, thoughtful responses
- Excellent at following complex, multi-step instructions
The catch
The Free plan has usage limits
Updated
ChatGPT
An assistant for everyday tasks
My take
It's an incredible workhorse, the desktop app is a pleasure to work with, and the usage quota on the Pro plan is quite something.
What it's good at
- Most versatile AI assistant — text, images, code, web browsing
- The GPT-6 models are excellent, reliable workhorses
The catch
Still less precise than Claude for the hardest reasoning and writing tasks
Updated
Gemini
Get quick answers and visual feedback
My take
Gemini isn't one of the strongest AI tools, and for real work I use ChatGPT Codex much more. But it's still my go-to for tiny work packages I need done quickly, like translating UI copy into other languages or getting UX feedback. The biggest plus for me is the near-instant answers.
What it's good at
- Near-instant answers with Gemini Flash
- Reads images and videos well
The catch
Not one of the strongest tools for bigger, demanding work
Updated
Gemini Notebook (formerly NotebookLM)
Understand your own documents
My take
The Audio Overview feature genuinely surprised me. Upload your research and it generates a natural-sounding podcast between two hosts talking through your content.
What it's good at
- 'Audio Overview' generates genuinely listenable podcast-style summaries
- Grounded in your own sources, which cuts down on hallucination
The catch
Audio Overview voices sound good but can be repetitive in format
Updated
Wispr Flow
Speak instead of typing, and take meeting notes
My take
I barely type anymore. Wispr Flow is good enough that I can whisper and it still understands me, in multiple languages, and it works on my phone and both my computers. It's one of the few tools that changed how I get through a day of writing. And its meeting note taker is excellent: in my tests it beat every other note taker I've used. Five stars.
What it's good at
- Incredibly accurate — works even when whispering
- Supports 100+ languages
The catch
Cloud-based processing raises privacy considerations
Updated
Suno
Make a personal song
My take
Suno is really good, and I've had a lot of fun making songs for my daughter and friends. A song about someone you know is a lovely reason to try AI music. I'd start with a few personal details and lyrics of your own, then try different musical styles.
What it's good at
- Creates complete songs with vocals and instrumentation from a brief
- You can supply your own lyrics and revise sections of the result
The catch
Generation credits and song downloads have separate limits
Updated
Higgsfield
Try image and video models
My take
I like having different models and creative controls in one place, so I can explore an idea without first committing to one model's way of doing things. I'd recommend it to someone who wants to try image and video generation and see what works for them.
What it's good at
- Lots of creative AI models to try in one place
- Combines image generation, video generation and editing tools
The catch
Generations, rerolls and upscaling can quickly use credits
Updated
Lovable
Turn an idea into an app
My take
Lovable is a great app builder, and I've built quite a few apps with it, even with a proper backend through Supabase. It is also great for quick prototyping and for figuring out difficult UX/UI questions.
What it's good at
- Generates complete working web apps from descriptions
- Real, exportable source code — not locked into the platform
The catch
Complex applications may need manual code refinement
Updated
Atlassian Rovo
Work with Jira and Confluence
My take
Rovo is the AI assistant for the Atlassian suite. We use Confluence for product specifications, and Rovo helps turn those requirements docs into epics and tickets in Jira. It doesn't feel really smart yet, but within a small, well-defined context, it's quite helpful. A solid 3 stars — useful but not impressive.
What it's good at
- Deep integration with Jira and Confluence
- Can create epics and tickets from requirements documents
The catch
Doesn't feel very smart yet — limited reasoning compared to standalone AI
Updated
Claude Academy & Documentation
Learn to get more from Claude
My take
Claude Academy is amazing. I look at it frequently and recommend it to all my coworkers. If you use Claude, this is where you start. What you won't find there anymore is a separate prompt engineering course, and I think that's right: a prompt should be written in the context of the whole system around it.
What it's good at
- Claude Academy covers everything from a first conversation to Claude Code, Cowork and agents
- Free, and kept up to date with the product
The catch
Naturally focused on Claude — you won't learn GPT-specific patterns here
Updated
Claude Code
Build and maintain software
My take
Claude Code is the tool that made me rethink what's possible with AI. I use it to build and maintain this entire website — writing components, managing content, deploying changes. It's like having a senior developer on hand who already knows the whole codebase.
What it's good at
- Understands your entire codebase in context
- Executes changes autonomously — not just suggestions
The catch
Not available on the Free plan: you need Pro, Max, Team, Enterprise, or API usage via a Console account
Updated
Claude Cowork
Delegate files and office tasks
My take
Cowork is how I think most knowledge work will happen within a year or two. Give it a folder of messy notes, come back to a structured report. Hand it expense screenshots, get a spreadsheet.
What it's good at
- Truly agentic: plans and executes multi-step tasks on its own
- Creates polished deliverables (Excel with formulas, PowerPoint, reports)
The catch
Requires a paid Claude plan (minimum $20/month Pro)
Updated
Claude for Excel
Work on spreadsheets with Claude
My take
Anthropic built a native add-in for Excel, allowing the Claude models to interact on code level with your worksheet. This allows you to use Claude as your highly proficient assistant for any spreadsheet task. Only downside I've seen so far: it costs a lot of tokens and uses up the limit of the lower-tier subscriptions quickly.
What it's good at
- Works at the code level inside the workbook you have open
- Explains how a number came about, with links to the cells it used
The catch
Consumes tokens quickly, especially with large worksheets
Updated
Claude for PowerPoint
Draft and edit PowerPoint slides
My take
Anthropic built a native add-in for PowerPoint, so Claude works on your deck at the code level instead of just talking about it. I've used it to draft slides from an outline and restructure a deck by describing what I want, and it saves real time on the fiddly parts. Only downside I've seen so far: it costs a lot of tokens and uses up the limit of the lower-tier subscriptions quickly.
What it's good at
- Builds and edits slides in your template: slide master, layouts, fonts and colours
- Edits one slide without regenerating the whole deck
The catch
Consumes tokens quickly, especially with large presentations
Updated
Claude for Word
Edit and format Word documents
My take
I don't work in Word much, but I played around with it a bit, and it's super useful for people who never fully learned Word or didn't keep up with all its changes and deeper features. Chat is now the primary interface I'd use for formatting my Word documents.
What it's good at
- Formatting by conversation: describe the result instead of hunting for the feature
- Edits keep the document's styles, numbering and formatting
The catch
Draws on the same usage limits as the rest of your Claude account
Updated
Claude in Chrome
Let an agent work in your browser
My take
I've used it to extract data and to control web apps for me. At Carewell I've used it to create and update the product documentation of our web app and to run tests on it. I even had Claude talk to our customer service chatbot: AI testing AI. Like the Office add-ins, it isn't really connected to my usual work setup, and most agent tools (Cowork, Claude Code, Codex, Gemini and others) have their own browser. So tech-savvy users will probably go that way. But it has its charm, because it's so simple.
What it's good at
- Nothing to set up: install the extension, and Claude works in the browser you already use
- Works on pages you're signed in to, including internal tools and old portals that have no connector
The catch
Hidden instructions on a web page can steer it, and Anthropic itself calls it risky
Updated
Copilot
An assistant inside Microsoft 365
My take
It's not as configurable and personalizable and doesn't learn as cleverly as ChatGPT and the likes. It doesn't interact cleverly with other software, amazingly not even with Microsoft products.
What it's good at
- Built into Windows and Office apps
- Enterprise data protection, also in free Copilot Chat for work accounts
The catch
Poor productization and UX
Updated
ElevenLabs
Turn writing into spoken audio
My take
ElevenLabs is brilliant at cloning voices, even to a professional level, and getting them to speak naturally across a lot of languages. Hand it a written dialogue and it'll voice it with multiple speakers, emotions, and audio effects. What still gives me pause is the same thing that makes it impressive: cloning a voice that convincingly is powerful, and easy to misuse.
What it's good at
- Most natural-sounding text-to-speech available
- Professional-grade voice cloning from the Creator plan up (Starter has instant cloning only), even for commercial use
The catch
Free tier very limited (10 min/month)
Updated
Figma Make
Explore designs and user flows
My take
Figma is quickly catching up with other UI generators, but it's not top-notch yet. I suspect that in the hands of a competent designer, it is much more powerful than in mine. However, it's really easy to create proper designs out of AI-generated screens, so I'm keeping an eye on this one.
What it's good at
- Deep integration with Figma's design ecosystem
- Generates designs aligned with your existing design system
The catch
Not yet top-notch compared to dedicated AI design tools
Updated
FLUX
Generate and edit images through an API
My take
I've only tested FLUX 3 a bit so far, but it seems to have caught up with the frontier again, both for generating images and for editing photos, and it's fairly fast while doing it. Quite impressive.
What it's good at
- FLUX 3 Image seems to have caught up with the frontier on generation and photo editing
- Fairly fast for its quality
The catch
Mostly an API product; most people will meet it inside other tools
Updated
Google AI Essentials
Get a structured introduction to AI
My take
The best 'send this to your colleague who hasn't started yet' resource I know. It's structured enough that people actually finish it, and the Google brand makes it easier to get approved as professional development. Not for power users.
What it's good at
- Clear, beginner-friendly structure — no prior AI knowledge needed
- Covers both concepts and practical application
The catch
Moves slowly for anyone who's already using AI tools daily
Updated
Google Antigravity
Build apps and work on your computer
My take
Antigravity is a software development workbench that can also be used for knowledge work, just like Claude Code. Since the 2.0 update it looks almost exactly like the ChatGPT and Claude desktop apps, and I enjoy using it to experiment with websites.
What it's good at
- Fully autonomous agents that plan, execute, test, and refactor
- Native desktop app (2.0) that feels like the ChatGPT and Claude apps
The catch
The strongest models need a paid Google AI plan
Updated
Google Jules
Run scheduled checks on a code repo
My take
These days I only use Jules as a monitoring tool: it has a look at my web projects and sends me an email from time to time. That still works. But I've mostly lost interest, and I think so has Google: the changelog has been quiet since March, and Google's coding energy has gone into Antigravity.
What it's good at
- Fully browser-based — no installation, no terminal, works on mobile
- Async workflow means you file a task and come back to a result
The catch
Async-only — no interactive back-and-forth while it works
Updated
GPT-Image-2.5
Create and edit images in ChatGPT
My take
What won me over first was the editing, being able to point at one part of an image and change just that. What keeps me here is how well it works out what I'm actually trying to visualize. I enjoy it for creative work as much as for the unglamorous editing that goes into UI design.
What it's good at
- Built directly into ChatGPT for zero-friction conversation
- Precise image editing allows clean modification of specific areas
The catch
Less artistic stylization than Midjourney
Updated
Grok Bot
A coding agent for technical users
My take
This is clearly an attempt at an alternative to Claude Code or OpenAI's Codex, and so far it's very minimalistic and built for techies. It's quite capable, though, which makes it my go-to when I've hit my token limits with the other models.
What it's good at
- Capable enough to pick up real work when your usual agent is out of quota
- Agents work in a persistent cloud VM, so they keep logins and files between sessions
The catch
Minimalistic and clearly built for technical users right now
Updated
Grok Imagine
Make images and video inside Grok
My take
Now, if you use Grok a lot, it's great to have this built in.
What it's good at
- Image 2.0 has caught up with the frontier image models in many respects
- Integrated directly into Grok and X for fast access
The catch
Images top out at 2K, where GPT-Image, Nano Banana and FLUX go up to 4K
Updated
Hermes Agent
Run an agent that builds reusable skills
My take
Typical agents treat every task as a blank slate, which gets annoying fast. Hermes has a built-in learning loop, so it actually builds and saves reusable skills as it works. If it figures out how to refactor a file or parse a document once, it remembers that pattern for the next task. It's model-agnostic too, so you aren't locked into a single provider — for a free, self-hosted framework, that pairing is rare.
What it's good at
- Built-in learning loop allows agent to create and refine reusable skills
- Persistent memory of user projects, files, and preferences across sessions
The catch
Setup is easier than it was but still more technical than a consumer app, with API keys and model endpoints to manage
Updated
Midjourney
Explore concept art and visual styles
My take
I used to experiment with Midjourney but never got more out of it than some playful edits. It's likely more suitable for design professionals who can invest the time in prompt craft. For non-designers, GPT-Image-2.5 or Nano Banana are more approachable starting points.
What it's good at
- Distinctive artistic aesthetic that still stands out
- V8.2 is the default model, with a focus on aesthetics and personalization
The catch
No free tier — paid plans only
Updated
n8n
Connect apps and automate workflows
My take
If your team has even a little technical ability, n8n beats Zapier and Make on both cost and flexibility.
What it's good at
- Self-hosted option means no per-execution fees and full data control
- Visual workflow builder makes complex automations accessible
The catch
Self-hosting requires some technical setup and maintenance
Updated
Nano Banana
A second opinion on images
My take
I use it as a second opinion: when I make something with GPT-Image-2.5, I run the same request through Nano Banana to compare.
What it's good at
- Good at a wide range of image jobs: photos, edits, mockups, infographics
- Up to 14 reference images to keep characters and objects consistent
The catch
Behind GPT-Image-2.5 as my first choice
Updated
OpenAI Academy
Learn to use ChatGPT with a purpose
My take
Solid starting point if ChatGPT is your primary tool. The structured courses beat random googling by a wide margin. Just don't stop here — the real learning happens when you apply these techniques to your own work.
What it's good at
- Well-structured courses organised by audience, from everyday work to building with Codex and the API
- Official source — always accurate for ChatGPT and GPT models
The catch
Some content feels more like product tutorials than deep learning
Updated
OpenAI Codex
Build software with an AI agent
My take
Codex is more or less on par with the latest Opus model in Claude Code, but it feels like it needs way more precise instructions, including what not to do, to yield the same results. Talking to Claude feels more natural, but depending on your tech stack, Codex might be the better, faster, and currently especially the cheaper choice.
What it's good at
- Excellent at writing, refactoring, and debugging production code
- Deep training on real pull requests and code review patterns
The catch
Requires more precise instructions than Claude Code for equivalent results
Updated
OpenClaw
Reach your AI assistant through chat
My take
OpenClaw is up to 4.5. The updates have been very good, and of the personal assistants I've tried, it's the only one I still use regularly. I like being able to ask it to do things through WhatsApp. The setup and permissions still need care, but it has stayed useful after the initial excitement wore off.
What it's good at
- The personal assistant I still use regularly after trying alternatives
- Very good updates have made it more useful to me over time
The catch
Setup, permissions and ongoing maintenance need technical attention
Updated
Perplexity
Research a question with sources
My take
I have no big research projects right now, so my subscription is on hold. But I'm not writing them off: if you research for a living or need sources you can really rely on, a dedicated research tool is still worth paying for. Four stars, still solid.
What it's good at
- Every answer comes with cited sources you can verify
- Deep Research is fast and stays close to its sources
The catch
Pro limits were cut more than once in 2026; heavy Deep Research use now pushes you toward the $200/month Max plan
Updated
Perplexity Computer
Research a topic and build an app
My take
I've lowered this to 3.5. It's too expensive for what I get out of it. I can still see the appeal if you're less technical, especially when you want it to research a topic and build an app to explore the results. That's the use case I'd consider it for.
What it's good at
- Combines web research, data analysis and app creation in one place
- Runs multi-step work in the cloud without a local development setup
The catch
Too expensive for the value I get from it
Updated
Poe.com
Use different AI models in one place
My take
All-in-one UI in one app and even accessible via API. For the price of a single subscription such as ChatGPT Pro. Minus points for bad user experience and UI that hasn't been improved at all in two years.
What it's good at
- Access to virtually every major AI model (GPT, Claude, Gemini, open source, and more)
- All modalities: text, image, video, audio in one app
The catch
User experience and UI haven't improved meaningfully in two years
Updated
Replit Agent
Build and publish an app in a browser
My take
Replit Agent is one of the most accessible ways to go from 'I have an idea' to 'here's a working app.' The zero-setup browser environment removes every excuse not to start building.
What it's good at
- Zero setup — everything runs in the browser, no local installs needed
- Agent 4 autonomously handles complex tasks: planning, debugging, testing, several tasks in parallel
The catch
Credits for the stronger Power and Max modes add up quickly in professional use
Updated
Replit Design
Try a clickable UX prototype
My take
I use Replit mostly for Replit Design. When I'm wrestling with a tricky UX concept, I describe what I want and get a working, clickable prototype in minutes.
What it's good at
- Generates interactive, clickable prototypes in minutes
- Working code — not just mockups, but real front-ends you can deploy
The catch
Turning a design into a new app needs a Core or Pro plan
Updated
Stitch
Try another approach to a UX problem
My take
Stitch is still powerful, but I've lowered it to 4. I realised it doesn't pull me back to keep working on designs. Designers may get more out of it than I do. For me, it's a useful place to get a second or third opinion on a UX question and see another way to solve it.
What it's good at
- Still a powerful way to explore interface ideas
- Useful for a second or third opinion on a UX question
The catch
It doesn't draw me back to continue working on designs
Updated
Superwhisper
Dictate privately on your own device
My take
Several of my colleagues use Superwhisper, and they still like it. Its strength is privacy: with the local models, your voice never leaves your device.
What it's good at
- Local models keep your voice on your device
- Custom modes with your choice of AI model for formatting and rewriting
The catch
Fully on-device only if you stick to the local models; the cloud models send audio out
Updated
UX Pilot
Turn requirements into UI screens
My take
I've lowered it to 4 because it hasn't kept up with other tools in my workflow. Moving between my company AI OS, where the context and specifications live, and a separate design tool is painful. It can still produce useful designs, but that friction matters.
What it's good at
- Helped me design a mobile app and many web screens without design expertise
- Generates wireframes, detailed UI screens and multi-screen flows
The catch
Hasn't kept up with other tools in my workflow
Updated
Veo 3.1
Animate images and create video clips
My take
It's just incredible what this video generator can do just from text prompts or one or a few reference images. To get to a proper video with consistent aesthetics, you still require an orchestrating tool like Google's Flow, though. I love it to animate images I've generated or to visualize something in an entertaining or humorous way.
What it's good at
- Incredible quality from text prompts alone
- Supports landscape and portrait aspect ratios
The catch
Limited to 4, 6 or 8 second clips
Updated
Zapier
Automate tasks without a server
My take
If you've never used workflow automation before, start here, not with n8n. Zapier removes every barrier: nothing to install, nothing to host, and the AI-assisted builder means you can describe what you want and have a working automation in ten minutes.
What it's good at
- No setup — sign up and start connecting apps within minutes
- 9,000+ apps
The catch
Per-task pricing gets expensive at volume
Updated
Stay in the loop
Don't miss what's next
One email when I've learned something worth sharing, roughly once a month.