How to Use ChatGPT for Design Ideas

ChatGPT is useful in design work before the visual stage: articulating your brief, exploring concept directions in words, and reasoning about why one approach fits the constraints better than another. Designers who use it early in a project move into visual development with sharper direction and fewer false starts. Those who use it without a clear brief get generic output that does not account for the real project.

The trap is session amnesia: each new conversation starts from zero. Client decisions, rejected directions and agreed constraints disappear unless you re-enter them. Tools like fenbs keep that context in one place outside any AI app, so the next session starts from the current brief rather than a blank slate.

Write your brief before asking for ideas

A brief that produces useful design ideas has at least four elements: what the work is for, who it is for, what constraints are non-negotiable (budget, medium, existing brand, timeline) and what the client explicitly wants to avoid. Give ChatGPT all four. A request that includes "our client sells sustainable outdoor gear to mid-30s professionals and wants to avoid anything that looks outdoorsy in a cliche way" will produce more useful directions than "give me branding ideas for an outdoor company."

You can also tell ChatGPT about past directions that did not work and why they did not work. "We tried a minimalist direction last round and the client said it felt too corporate" narrows the space in a useful way. ChatGPT cannot remember previous sessions unless you tell it what happened in them.

Developing concept directions in words

Ask for three distinct concept directions, not one idea. Three directions force a genuine choice. For each, ask ChatGPT to describe the core idea in one sentence, the visual language it implies (typeface quality, color temperature, texture, composition style), and one example from an adjacent field that captures the feeling. That level of specificity is enough to sketch from without constraining the visual judgment.

Once you have three directions, ask ChatGPT to argue for each one against your brief. "Why would a designer choose Direction A over Direction B for this client?" That argument surfaces constraints your brief implied but did not state explicitly. Often the strongest argument for a direction reveals a client need that was not in the original brief at all.

Evaluating ideas against real constraints

After choosing a direction to develop, use ChatGPT to stress-test it against the constraints you know. "This direction uses a lot of textured illustration. Where does that break down?" or "If the budget limits us to two colors, how does this concept hold up?" This kind of adversarial questioning finds the weak points before you invest hours in development.

For interior design ideas, ChatGPT for interior design covers how to apply the same constraint-driven approach to spatial and decorating work. For graphic design specifically, ChatGPT for graphic design goes deeper into type, color and composition prompting.

Presenting ideas to clients

ChatGPT can help write the narrative for a design presentation: the single sentence that frames each concept direction, the paragraph that explains the reasoning, and the questions you want the client to answer to help narrow the choice. Ask it to write these in the tone your studio uses, and edit the result before presenting.

After the client review, record what happened: which direction was chosen, which elements were approved as-is, which were modified, and which were ruled out entirely. "The client approved the typography direction but wants to see two options for the icon treatment" is a decision that shapes the next round. Without recording it, you start the next session without that constraint and risk re-presenting something the client already rejected.

What ChatGPT cannot do in design work

ChatGPT generates text. It can produce conceptual images using DALL-E in ChatGPT on supported plans, but those images are illustrative and imprecise: wrong proportions, inconsistent typefaces, low brand fidelity. They are useful for client mood conversations, not for production development. Final design work happens in your actual design tools.

ChatGPT also cannot look at your own past work, access brand guidelines from a file unless you upload them in the session, or compare your direction to live competitor work. It reasons from what you provide. The quality of the ideas you get back is directly proportional to the quality of your brief and context you put in.

How to use ChatGPT for design ideas with fenbs

A brief setup at the start of each design project removes the context problem and makes every session start from the current state of the work.

1. Create a free account. Start a project for the design engagement. Add your brief as the project description: what it is for, who it is for, the hard constraints and the explicit avoids.
2. Add tasks for each project phase: brief development, concept generation, client review and development. Note the current phase in Where we left off.
3. After each client review session, record the outcome as Decisions and rules: what was approved, what changed and what was ruled out. Date each entry and name who decided.
4. Add reference links to Links and listings: the client's existing brand pages, competitor examples the client mentioned and any specific products or references they approved.
5. At the start of each new working session, ask your assistant to read the project context and summarize the current brief and the decisions made so far before generating any new ideas.