Tools
A short list,and the reason it is short.
Four tools. One per category. Everything here is something I pay for out of my own account and use on work with a client or a deadline attached, which is the only test I trust and also the reason this list will never be comprehensive.
- Four picks · one per category
- Affiliate links marked and tagged rel="sponsored"
- No paid placement
Disclosure
Read this before the list, not after it.
A disclosure a reader meets after they have already decided is not a disclosure. So it goes here, in the loudest block on the page, ahead of every recommendation.
Some links below are affiliate links, marked as such. If you buy through one I may earn a commission at no extra cost to you. It does not affect the rating: I have left tools I was offered money to include off this list entirely.
In practice that means three things. Every affiliate link on this page is tagged rel="sponsored" in the markup, so a search engine sees exactly what you see. The commission does not change the order of the list or the number next to a tool. And nothing here is paid placement: if a vendor ever buys space on this site it will be labelled as an advertisement and it will not be on this list.
The reason to care is simple enough. An affiliate link gives me a reason to want you to buy, and the only useful defence against that is telling you before you read rather than after.
The list
The stack I actually pay for.
Short list on purpose. I review things I use on real work, and I say when something is not worth it: a directory where everything scores well is an advert, not a review.
[PLACEHOLDER: Tool 1 name]
[PLACEHOLDER: category, e.g. Coding]
[PLACEHOLDER: rating out of 5]
[PLACEHOLDER: Your one-line verdict]
Use it when — [PLACEHOLDER: The specific situation this is the right pick for]
[PLACEHOLDER: Tool 2 name]
[PLACEHOLDER: category, e.g. Writing]
[PLACEHOLDER: rating out of 5]
[PLACEHOLDER: Your one-line verdict]
Use it when — [PLACEHOLDER: The specific situation this is the right pick for]
[PLACEHOLDER: Tool 3 name]
[PLACEHOLDER: category, e.g. Images]
[PLACEHOLDER: rating out of 5]
[PLACEHOLDER: Your one-line verdict]
Use it when — [PLACEHOLDER: The specific situation this is the right pick for]
[PLACEHOLDER: Tool 4 name]
[PLACEHOLDER: category, e.g. Automation]
[PLACEHOLDER: rating out of 5]
[PLACEHOLDER: Your one-line verdict]
Use it when — [PLACEHOLDER: The specific situation this is the right pick for]
Categories
Four jobs, one pick each.
The list is organised by the job, not by the vendor. One tool per job, because the second-best tool in a category is a tab you open twice and then stop opening.
01
Coding
The tool the brief goes into. It matters more than the other three put together, and the gap between a good pick and a bad one is measured in whole afternoons.
02
Writing
Drafting, editing, and the unglamorous work of turning a rough argument into something a stranger can follow. Judged on how much of the draft survives contact with me.
03
Images
Thumbnails, diagrams and the visual work that used to mean waiting on a designer for something that was never worth a designer's time in the first place.
04
Automation
The glue. Whatever runs the job on a schedule, so it stops living in a browser tab and starts living somewhere that does not need me awake.
A fifth category appears when something in it earns a slot, and not before. Empty is a legitimate answer.
Method
How I rate, and what the number means.
The rating is one judgement, not a weighted average of features. The question behind it is whether I would pay for it again this month, knowing what I know now.
01
I paid for it
Out of my own account, at the normal price, for at least a month. Trials and free tiers do not count, because the interesting failures arrive after the novelty and after the invoice.
02
It survived a real deadline
Not a demo and not a weekend project. Something with a client, a launch or a publication attached, where the cost of the tool being wrong was mine to carry.
03
It does one job better than the alternative
Breadth is easy to market and hard to use. A tool earns its place on the thing it is best at, and the write-up names that thing rather than repeating the feature page.
04
Its failure mode is survivable
All of these get things wrong. What separates them is whether they are obviously wrong or quietly wrong. Quietly wrong costs far more, and it is the usual reason something scores lower than it reads.
05
The pricing model matches the work
Per seat, per run, per token. The shape of the bill has to match the shape of the job, because a tool that is cheap right up until it is useful is not cheap.
What the number means
- 5: I would buy it again today without opening the invoice.
- 4: It earns the money, and the write-up names the limitation.
- 3: I still pay for it, and I would switch tomorrow if something better arrived.
- Below 3: It does not appear here at all. A tool I have stopped paying for is an anecdote, not a recommendation.
There are no half points, no weighted scoring and no category-winner badge. Those exist to make a list look rigorous. One number and a sentence of reasoning is more honest, and much easier for you to argue with.
Why it is short
Four is not a gap in the research.
A directory with two hundred entries is a different product from a recommendation. It is built to rank, it is maintained by nobody in particular, and everything in it scores between four and five, because the entries that scored two were quietly dropped for not converting.
This list is short because it only holds tools I am currently paying for. That is a much smaller set than the tools I have tried, and a far smaller set than the tools I could write a paragraph about. The constraint is the point. If you want the wider view, the newsletter covers what shipped this week, including the things that are not worth buying.
The honest limit: this is one operator's stack, built around specification-driven work with AI, on the kind of projects I take. It is not a survey and nothing here is benchmarked. If your bottleneck is different from mine, your right answer may be different too, and a list of four cannot tell you that.
Removal
What gets a tool taken off this list.
A list that only ever grows is a list nobody is editing. Any one of these is enough on its own.
- The price moves and the value does not. A rise is fine. A rise that arrives with a new usage cap is a different product at a higher price.
- It gets worse at the job I bought it for. Models get swapped and defaults get changed, and the thing that justified the subscription quietly leaves.
- Something free does the same job well enough that I would not pay again. That is a removal, not a lower score.
- A month goes by and I have not opened it. Usage is the one vote an affiliate link cannot influence.
- The company changes what it does with your data, or makes it hard to find out. Training on client work is a removal by itself.
- It breaks and nobody answers. Support is part of what you are buying, and you only ever find out at the worst possible moment.
When something comes off, the entry says so and says why rather than disappearing overnight. A removed tool is usually more useful to read about than a kept one, and it is the only part of a list like this that cannot be faked.
Before you buy anything
The awkward questions about lists like this.
Do you get paid if I click these?
Why is the tool everyone is talking about not on here?
Do I need to buy these to do the thing you teach?
How often does this change?
Can I suggest a tool?
While you are here
The tool is not the bottleneck. The brief is.
Every tool on this page produces better work from a specification than from an instruction, and that is a skill rather than a purchase. The free course teaches the version I use. The call is thirty minutes and there is no pitch in it.
P.S. If you are about to buy something off this page, buy the one that removes a bottleneck you hit this week. Not the one with the highest number next to it.