AI Tools7 min read

How to Make Diagrams With ChatGPT (2026): Flowcharts, Timelines, Mind Maps

The two-step method I use to get usable diagrams out of ChatGPT: ask which diagram type fits first, then generate with specifics. Updated for 2026: the Diagrams Show Me GPT (plugins are gone), Mermaid code you can edit anywhere, and when to use image generation instead.

Choosing the right diagram type matters more than the diagram itself. That one shift in thinking is what took my diagrams from generic to genuinely useful, and it's what separates people who get value from AI diagram tools from people who just end up with a confusing flowchart they never use.

I tested more than 10 ChatGPT diagram tools before settling on one: Diagrams: Show Me. Whimsical, Diagram Genius, UML Diagram Creation Expert, Diagrams Data Research, I ran them all. Diagrams: Show Me consistently gave the most usable output and the cleanest hand-off to Miro for editing. Not sponsored, just the one that worked.

What changed since I recorded this#

• Plugins are gone; GPTs replaced them. The video shows Diagrams: Show Me as a ChatGPT plugin. OpenAI retired plugins in 2024, and the same tool now lives in the GPT Store as a GPT, where it remains the most-used diagram option alongside Whimsical Diagrams, Mermaid Chart, and Lucid (justainews, The Bricks). Search the GPT Store for "Diagrams: Show Me," open it, and everything below works the same.
• Mermaid is the portable layer. Under the hood these GPTs write Mermaid code. Any modern ChatGPT model will write Mermaid directly if you ask, and you can paste it into Mermaid Chart, draw.io, Notion, GitHub, or Obsidian and edit it there. Mermaid's official GPT is documented at mermaid.ai.
• Image generation is now an option, with a catch. ChatGPT can draw a diagram as a picture. It's good for a slide-ready illustration and bad for anything you need to edit or where every label must be exact; check the text before you use it. For process diagrams I still go Mermaid first.
• Plans: the GPT route works on the free plan. Image generation is limited on Free and expands on Go and Plus; OpenAI's pricing page has the current allowances.

Start with the right diagram type#

Before you generate anything, ask ChatGPT what you should generate. This sounds obvious but almost nobody does it.

The logic is simple. Different diagram types serve completely different purposes:

• Flowchart or timeline, if you're showing a process or sequence of events
• Mind map, if you're brainstorming or organizing a cluster of related ideas
• Comparison chart or bar graph, if you're weighing options or highlighting advantages

Picking the wrong type doesn't just look bad, it actively obscures the point you're trying to make. A mind map of a step-by-step process is confusing. A flowchart of loosely related concepts is cluttered. The diagram type is load-bearing.

The prompt that does the heavy lifting#

Here's the exact prompt structure I use before creating anything:

"Please give me some suggestions on what diagrams I could use to explain my article better. My article is below: [paste content]"

I did this with a blog post I'd written about stock appreciation rights (SARs), a topic that's dense and text-heavy with no visual content. ChatGPT came back recommending a flowchart, a timeline, a comparison chart, a bar or line graph, and a mind map. From there, I picked the one that fit what I was trying to communicate and asked it to generate that specific diagram. The quality of that first prompt matters a lot, and the six-slot prompt template I use for most tasks applies here too: name the format, the audience, and the constraints up front.

That two-step approach, ask first, generate second, consistently produces better output than jumping straight to "make me a flowchart about X."

If you're creating diagrams to support blog content and want to make sure you're writing about topics worth visualizing in the first place, my SEO Keyword Generator GPT can help you find the angles that actually get search traffic before you build anything around them.

Walking through a real session#

Flowchart first. After pasting my SAR blog post and getting diagram recommendations, I asked for a flowchart. The output mapped the full SAR lifecycle: grant, vesting period, condition met or not, exercise window, payout in cash or stock. Solid structure, immediately readable. From there, Diagrams: Show Me gives you options to view the full diagram, download a PNG, or open it in an editor.

Then a timeline. For this one, I got specific. Vague prompts produce vague diagrams. I told ChatGPT exactly what milestones to include: grant date, three-year vesting period from 2022 to 2024, exercise window from 2025 to 2028. The more concrete the input, the more accurate the output. If you want a timeline that actually represents something, give it the actual dates.

Then a mind map. I asked for a mind map of the benefits of SARs for employees and companies. The initial output was, honestly, a mess: nodes scattered, layout hard to follow. This is normal. The generated diagram is a starting point, not a finished product.

Where Miro and Mermaid come in#

ChatGPT is a good starting point, and after that you make the chart exactly how you want it. That's the whole workflow in one sentence. The GPT gets you 70 percent of the way there. An editor gets you the rest.

When you open the diagram in Miro, it imports directly. From there you can rearrange nodes, adjust styling, bold or highlight specific text, and shuffle the layout until it actually looks like something you'd put in a presentation or on a page. No design skills required: drag, click, done.

If you prefer code-based editing, take the Mermaid code instead. Paste it into Mermaid Chart or draw.io, switch the theme (dark mode changes the whole feel), and export SVG or PNG. Because the diagram is text, you can also version it, drop it into a README, or have ChatGPT revise one node without regenerating everything.

The diagrams I ended up with after editing: a SAR versus stock options comparison chart, a benefits breakdown for employees versus companies, and a clean lifecycle timeline showing how SARs are granted and exercised. None of them looked like the raw output. All of them were usable.

The actual workflow, condensed#

1. Paste your content and ask ChatGPT which diagram type fits best
2. Pick the recommendation that matches what you're trying to show
3. Prompt with specifics: dates, milestones, named options, whatever the diagram needs to be accurate
4. Generate through the Diagrams: Show Me GPT, or ask for Mermaid code directly
5. Open in Miro or a Mermaid editor and clean up the layout and styling
6. Export as PNG or SVG

One ChatGPT session. No Illustrator, no Figma, no design background. If you want to go deeper on building custom AI tools around your content process, building your own Custom GPT is a natural next step, and if you're working with data you've already got in spreadsheets, turning an Excel file into a live dashboard website with ChatGPT picks up where a static chart stops.

The GPT ecosystem changes fast. Try a few, see what gives you the cleanest output for your use case. But start with Diagrams: Show Me, or with plain Mermaid if you'd rather own the file.

Watch the full video on YouTube: https://youtu.be/jXBvHBgjugQ

This post contains affiliate links. I only recommend tools I actually use.