Six Of These Nine Ladders Go Up By Swapping The Tool. Three Go Up By Keeping It.

Eleven charts, one swipe. Each one takes a job — research, writing, design, automation — and sorts tools into three steps: Beginners, Intermediate, Professionals. The caption asks which ranking you agree with most, which is a fair question and not the interesting one.

The interesting thing only shows up when you stop reading them one at a time and lay them side by side. Do that and the charts stop being nine opinions about nine jobs and start being one argument about what a professional tool is — an argument the author seems to half-know they are making, because on three of the charts they drop the product names entirely.

Below: all nine charts I could read, the pattern that runs across them, three names that need updating, and the line of small print that is better than most of the cards it sits under.

All nine charts, side by side

Here they are as a single table. Read down the Professionals column first.

Six ladders swap the tool. Three keep it.

On six of the nine charts, going from intermediate to professional means picking up something new. Gemini gives way to ChatGPT gives way to Claude. Cursor gives way to Claude Code. Suno gives way to ElevenLabs. You climb by buying, installing or signing up for a different product.

On three charts, that stops happening. For Writing, the top step is Claude or ChatGPT — the same two names that were already on the lower steps — plus custom skills. For Presentations, it is Claude plus presentation skills. For Design, it is Claude plus design skills, and the middle step has already done the same trick: ChatGPT plus image skills plus reference images.

So on the Design chart, two of the three steps are not products at all. They are configurations. The thing that moves you up the chart is not what you are holding; it is what you have written down and handed to it.

And this is not vague. A skill, in the Claude sense, is a specific thing on disk: a SKILL.md file with YAML frontmatter and instructions, invoked by typing a slash and its name, or loaded automatically when it is relevant. Custom commands were merged into the same mechanism, so a file at .claude/commands/deploy.md and a skill at .claude/skills/deploy/SKILL.md both give you /deploy. When the card says “+ design skills,” it is pointing at a real, nameable artefact you can write once and reuse. If the vocabulary around this is still fuzzy, start here.

The same tool keeps changing steps

Now read the table the other way: not by job, but by tool.

ChatGPT is the clearest case. It is a beginner tool for research and for writing, an intermediate tool for brainstorming and design, and the professional pick for image generation. That is not an inconsistency the author failed to notice. It is the chart telling you something true: the step a tool sits on is a property of the job, not of the tool.

Which makes the three “+ skills” rows the honest summary of the whole set. On the jobs where the author clearly has the most hands-on experience — writing, presentations, design — they stopped naming products and started naming preparation. Picking the best tool for each job is a real exercise, but it has a ceiling, and these three rows are where the card hits it.

And the automation chart does it too, in disguise

Look again at the Professionals step for Automating Tasks: OpenClaw and Hermes Agent. On the face of it that is a product swap — Make and Zapier out, something more serious in. It is not.

OpenClaw is open source, MIT-licensed and self-hosted. It began as Clawdbot in 2025, was renamed Moltbot in early 2026 over a trademark, and became OpenClaw in January 2026. It does not come with a model; you point it at Claude, at a GPT model, or at something running locally. Hermes Agent, from Nous Research, shipped in February 2026 on the same terms: MIT, self-hosted, you pay for your own tokens and hosting.

So the step up is not from a cheap product to an expensive one. It is from a hosted service to infrastructure you run, driven by the same models you were already using. Same pattern as the writing chart, different clothes.

And then Hermes Agent closes the loop completely, because its headline feature is a learning loop that turns completed tasks into reusable skills and carries them into later sessions. The professional tier of the automation chart is a program whose selling point is that it accumulates skills. The card could not have made the argument more plainly if it had tried.

One thing the card does not mention, and should: OpenClaw’s security posture has been a documented problem, with large numbers of instances left exposed on the open internet, and in March 2026 Chinese authorities restricted its use on state enterprise and government machines. If you put it on a work laptop, read its hardening guide first.

Three names on the Professionals row

Because the card asks you to go and use these, I checked every tool named across the nine charts. Most are fine. Three are not.

None of this makes the Research or Voice charts wrong, exactly. Gemini Notebook is the same tool it was under its old name, and if the card means a different Voicebox then the card simply does not say which. But a chart whose whole job is to tell a beginner what to go and install is a chart where the names have to be current, and the top step of the Research chart currently sends you to two products you will not find under the names printed.

There is a smaller mismatch worth naming too. Vicena is a laboratory and engineering research assistant — literature review, experimental data, simulations, chemistry and semiconductors. The audience for a nine-card AI tools carousel is mostly people doing market research, competitor research and content research. That is a different job with the same word attached to it. Checking the label against what the thing actually is is the step most tool charts skip.

What the card gets right

The line of small print under every one of the nine charts is better than the charts: one personal view of tools by level, not a ranking, with a reminder that tools and features change fast, so test them on your own work and check each official site.

That is three concessions in one sentence — it is personal, it is not a ranking, and it will go stale — and it pre-empts most of the fact-checking above. The author told you the names would drift before the names drifted. Most cards of this shape say nothing of the kind, and the ones that do usually bury it.

The structure is also better than a flat list. Sorting by job rather than by tool is the right axis, because it is the axis on which the answer actually changes. And putting three steps on each chart, rather than one winner, is what made the “+ skills” pattern visible at all. A ranking could not have shown that.

Four honest caveats

1. I read nine of eleven charts. The carousel holds eleven images and Facebook shows five in the grid with the rest behind a login; I recovered four more from the page’s photo tab, which leaves two I have not seen. Every count in this post — six versus three, ChatGPT on all three steps — is a count over the nine I could read. Two more charts could shift the ratio. They are unlikely to remove the three “+ skills” rows, which are the argument.
2. The grouping is mine. The card never says “six swap the tool, three keep it.” It presents nine independent opinions and explicitly calls them a personal view rather than a ranking. Reading them as one argument is my reading, and someone could fairly call the Vibe Coding chart a fourth “keep it” case on the grounds that Claude Code is the same model in a different wrapper.
3. A redirect is evidence, not proof. mirrorthink.ai resolving to vicena.ai is strong evidence the two are connected, and Vicena is in the same business. But Vicena’s own site does not mention MirrorThink, so I cannot tell you whether that was a rename, an acquisition or a domain changing hands.
4. “Voicebox” is a common word. I have matched it to Meta’s unreleased 2023 speech model because that is by far the best-known thing with the name, and because it sits next to ElevenLabs on a professional voice row. If the card means a smaller product with the same name, my objection evaporates and the card simply needed a link.

The one-line version

Six of these nine charts say you get better by changing what you hold; three of them — the three on the jobs the author plainly knows best — say you get better by keeping what you hold and writing down how you want it used, which is also why the same tool keeps showing up on three different steps.