How I Solved My Image and Infographic Headache With ChatGPT

Here is the real story.

I would be on my phone, mid-idea, using the ChatGPT app to talk through a strategy, article, service concept or campaign angle.

Then the obvious next thought would appear.

This needs an infographic. Or a feature image. Or a social carousel. Or a visual explaining the process.

ChatGPT could generate the image there and then, which is useful. OpenAI’s own help material confirms that ChatGPT can create and edit images inside the product, including generating images from descriptions and editing images through the image interface.

But that did not solve my actual problem.

The problem was workflow.

I did not want to download images on my phone. I did not want files saved in the wrong place. I did not want to lose track of which image belonged to which idea. I did not want to regenerate an image later because I forgot to save it properly.

I needed a better system.

The real issue: creative flow kept getting interrupted

When an idea is forming, stopping to create, download, rename and store an image can break the flow.

That is especially true when the idea appears during a voice conversation on mobile.

Maybe I am walking. Maybe I am cooking. Maybe I am thinking through a blog post while away from the desk. Maybe I am developing a service page, podcast topic or LinkedIn post and suddenly realise it needs a visual asset.

At that moment, I do not want to become a file manager. I want to keep thinking.

Before I built the system, the options were messy:

• Generate the image immediately and risk losing it later.
• Download it to the wrong device or folder.
• Make a vague note and hope I understood it later.
• Ask ChatGPT to regenerate it when I returned to desktop.
• Break the strategy conversation to deal with production admin.

None of those were ideal.

The whole point of using generative AI is to reduce friction, not create another layer of creative admin.

The breakthrough: the Imagery Backlog

The solution was simple.

Instead of generating every image the moment the idea appeared, I created an Imagery Backlog.

Now, when a visual idea comes up, I say:

That is it.

The idea is logged. The visual is queued. I keep working.

Later, when I am on desktop and ready to generate assets properly, I open the backlog and say:

Now the image is created at the right time, on the right device, in the right workflow.

How the Imagery Backlog works

The workflow is deliberately simple. It has four stages.

Example commands

• “Add to Imagery Backlog: hero image for an article about AI fatigue.”
• “Add to Imagery Backlog: infographic explaining capability debt.”
• “Add to Imagery Backlog: LinkedIn carousel showing how hidden work becomes an AI use case.”
• “Add to Imagery Backlog: visual for AI agents as a digital leadership team.”

The structure of the backlog

The Imagery Backlog is not complicated. It is a simple operating list.

The exact columns can change depending on the business, but the principle is the same. Every visual idea needs a number, a description, a status and a clear link to the work it supports.

Why this matters for content workflows

Visual content is often where good workflows fall apart.

Articles get written but never get a proper feature image. Workshop ideas need diagrams that never get created. Social posts would be stronger with a carousel, but the visual is too much effort. Service pages need a simple explanatory graphic, but it gets pushed aside.

The problem is rarely lack of intent.

The problem is that visual production is often treated as a separate task, disconnected from the thinking that created the need for the visual in the first place.

The Imagery Backlog fixes that.

It captures the visual requirement at the moment it appears, while the idea is still fresh.

This is a practical example of workflow automation thinking, even if the first version is manual. The goal is to reduce handoff loss, double handling and forgotten work.

Why this is better than generating everything immediately

Generating everything immediately feels productive, but it can create problems.

• The image may not have enough context yet.
• The idea may change while the article or page develops.
• The image may be saved in the wrong place.
• The wrong size or format may be generated.
• The visual may need brand alignment that is easier to apply later on desktop.

By using a backlog, I separate two different activities:

• Creative capture: saving the visual idea when it appears.
• Asset production: generating the image when the context, use case and storage location are clear.

That separation makes the workflow calmer. It also improves quality.

How this connects to ChatGPT Projects

For recurring workflows, this type of backlog works best inside a dedicated ChatGPT Project or a clearly named recurring chat.

OpenAI describes Projects in ChatGPT as workspaces that can keep chats, files and instructions together for longer-running work.

That matters because an Imagery Backlog is not a one-off prompt. It is a working system.

It needs to remember structure, naming conventions, status fields and the relationship between ideas and visual assets.

A project-based setup can hold:

• The Imagery Backlog Index.
• Brand image rules.
• Preferred dimensions.
• Visual style guidance.
• Completed asset notes.
• Future image requests.

For Changeable, this type of workflow is especially useful because the site uses a consistent professional style across articles, service pages and case studies.

It also supports related brands like Zero to AI and Ministry of Insights, where imagery needs to fit different audiences and visual styles.

How this helps with brand consistency

AI image generation becomes more useful when it is guided by a style system.

Without a style system, every image can look like it belongs to a different brand.

The Imagery Backlog can include brand notes such as:

• Preferred colour palette.
• Image style.
• Composition rules.
• Audience considerations.
• Safe-zone requirements.
• Export dimensions.
• Use case.
• Do-not-use visual clichés.

How SMEs can use this

Small businesses often create content in bursts.

A business owner might write a blog post, think of a social image, need a workshop diagram, want a product graphic and then get pulled into client work before any of it gets finished.

An Imagery Backlog gives that business a simple way to keep track of creative production.

It can be used for:

• Blog feature images.
• Social media carousels.
• Infographics.
• Website section visuals.
• Email campaign graphics.
• Workshop diagrams.
• Lead magnet graphics.
• Product explainers.
• Proposal visuals.
• Internal training images.

The value is not only image creation. The value is not losing ideas between the moment they appear and the moment they can be produced properly.

How councils and larger organisations could use it

The same workflow can also help councils, public sector teams and larger organisations.

These teams often need visuals for:

• Public consultation material.
• Process explainers.
• Internal change communications.
• Strategy documents.
• Community engagement updates.
• Training resources.
• Policy explainers.
• Workshop facilitation.
• Board or committee reports.
• Service improvement diagrams.

In those environments, governance matters more.

The backlog should include review steps, accessibility checks, brand approval, privacy considerations and human sign-off before publication.

This is where AI governance becomes practical. It is not about stopping people from using AI. It is about making sure outputs are reviewed, accurate, appropriate and safe for the intended audience.

Accessibility should be part of the workflow

Infographics can be useful, but they can also create accessibility issues if they are not designed carefully.

A strong imagery workflow should include accessibility checks.

• Use readable text sizes.
• Avoid low-contrast colour combinations.
• Do not rely only on colour to communicate meaning.
• Provide alt text for images.
• Include a text-based explanation where needed.
• Keep diagrams simple enough to understand.
• Check mobile readability before publishing.

The W3C Web Accessibility Initiative provides guidance on web content accessibility, including making content perceivable, understandable and robust for different users.

For any business using AI-generated infographics, accessibility should be designed into the workflow rather than added at the end.

Prompt quality still matters

The Imagery Backlog solves the workflow problem, but image quality still depends on good direction.

A weak prompt creates weak output.

A useful visual prompt should include:

• What the image is for.
• The target audience.
• The format and dimensions.
• The main message.
• The visual style.
• The mood.
• Any brand colours or design rules.
• What to avoid.

That is far stronger than:

This is where AI becomes more useful when paired with human creative direction.

Why this solved the headache

The headache was not that images were impossible to create.

The headache was that visual ideas were arriving in the wrong place at the wrong time.

The Imagery Backlog fixed that by giving every visual idea somewhere to go.

Now the process is clean:

• The idea appears.
• I add it to the backlog.
• The backlog gives it structure.
• I continue the original work.
• The image is generated later.
• The completed item is marked off.

That is a small system, but it removes a lot of friction.

It also shows why AI productivity is often less about one dramatic automation and more about fixing the small workflow breaks that happen every day.

How this connects to hidden work

Before this system, image management was hidden work.

It was not listed as a formal task. It did not appear in a process map. But it still consumed energy.

Remembering the image idea, finding the old chat, regenerating a visual, downloading it on the wrong device, searching for the right file and recreating the prompt all added unnecessary drag.

The Imagery Backlog made that hidden work visible.

Once it was visible, it became easy to redesign.

This is the same principle behind finding hidden work and shadow processes with AI. The best automation opportunities often sit inside small repeated frustrations that nobody has formally named.

How to build your own Imagery Backlog

You can build a simple version without any special tools.

A simple reusable prompt

Here is a basic prompt you can adapt.

This small instruction turns ChatGPT from a one-off image generator into a lightweight creative operations assistant.

What this teaches about AI workflow design

The Imagery Backlog is not only about images.

It is a lesson in AI workflow design.

AI becomes far more useful when it supports a repeatable process.

That process should be:

• Easy to trigger.
• Structured enough to retrieve later.
• Clear about status.
• Connected to the wider work.
• Reviewed by a human.
• Simple enough to keep using.

This same logic can be applied to ideas, risks, blog topics, workshop assets, client questions, sales objections, process issues and automation opportunities.

That is why Changeable focuses on practical AI strategy and workflow fit rather than AI tool use in isolation.

What Changeable helps with

Changeable helps New Zealand businesses use AI to improve real workflows, not just generate isolated outputs.

About the author

Steve Wilson is the founder of Changeable and Ministry of Insights, providing AI strategy, governance and automation consulting for organisations navigating the gap between AI ambition and operational reality.

For people and teams still building confidence with AI before implementation, visit Zero to AI.