Brainstorm

Podcast 2 — In conversation

Speaker A: Our team spent three hours in a brainstorm yesterday and came out with more ideas than we expected. Though I have to say the first hour was a bit slow before people relaxed into it.

Speaker B: That is actually the most common pattern. The first phase of a brainstorm tends to produce obvious ideas that people were already thinking before the session started. The more interesting ideas usually come in the second half, once the obvious territory has been cleared and people start making less predictable connections.

Speaker A: Right. We found that was true. The idea we ended up pursuing came quite late in the session. Someone made a connection between two earlier ideas that had seemed unrelated.

Speaker B: That is precisely what the no-judgment rule is designed to allow. If people are evaluating as they go, they prune ideas before the connections have a chance to form. Osborn's original insight was that criticism kills generativity, so you have to separate the two phases completely.

Speaker A: How does brainstorm differ from ideation? I see that word used a lot in design contexts.

Speaker B: Ideation is the broader design-thinking term for the entire idea-generation phase of a process. A brainstorm is one specific technique within ideation. You can ideate through brainstorming, but also through mind mapping, sketching, card sorting, competitive analysis, or silent writing methods. Brainstorm implies a group verbal session. Ideation encompasses a wider range of methods.

Speaker A: What about brainwave? We mentioned it as distinct from brainchild.

Speaker B: A brainwave is the individual sudden flash of inspiration, the aha moment. A brainstorm is the organised group method for generating many ideas. You might have a brainwave during a brainstorm, or a brainwave might come to you alone in the bath. They are related but operate at different scales: brainwave is solitary and spontaneous; brainstorm is collective and structured.

Speaker A: A mistake I see is running a brainstorm where someone immediately evaluates or dismisses ideas as they come up. That kills the whole purpose.

Speaker B: That is the single most common failure mode. The moment someone says that will never work in the generation phase, the room shuts down. People stop offering half-formed ideas because they fear judgment, and the half-formed ideas are often the most valuable raw material. The evaluation phase must be completely separate, temporally and even physically if possible, from the generation phase. That separation is not a soft preference; it is the methodological core of what makes brainstorming work.

Podcast 3 — Prompt engineering

Speaker B: Brainstorm is one of the most versatile and productively specific words in the AI prompt engineering vocabulary. It signals to the model that you want divergent, generative, high-volume output rather than a single definitive answer. It shapes the response style, the quantity of suggestions, and the creative register of what comes back. Used deliberately, it dramatically improves ideation prompts for developers building creative tools, project planning systems, and collaborative platforms.

Speaker A: So adding brainstorm to a prompt changes not just the content but the format of the model's response?

Speaker B: Significantly. It signals quantity over selection, breadth over depth, and creative range over conservative recommendation. Here is a feature ideation prompt for a development team:

Speaker B: Brainstorm twenty distinct feature ideas for a freelancer invoice management platform. For each feature, give a one-sentence description, the primary user benefit, and a rough implementation complexity rating of low, medium, or high. Cover UI improvements, automation opportunities, client communication tools, payment integrations, and reporting capabilities. Do not filter for feasibility yet; quantity and variety matter most at this stage.

Speaker A: The explicit instruction not to filter for feasibility at this stage is exactly the no-judgment rule from classic brainstorming methodology applied directly to a model prompt.

Speaker B: And the model responds with genuine breadth rather than the top five safe options. Now a database schema ideation prompt:

Speaker B: Brainstorm eight alternative database schema approaches for a multi-tenant SaaS HR platform. For each approach describe the tenancy isolation strategy, whether it uses separate databases per tenant, a shared database with a tenant_id column, or a hybrid approach, the key trade-offs for scalability and data isolation, and a one-sentence recommendation on which team size and scale the approach suits best.

Speaker A: Getting eight schema approaches with explicit trade-offs at the brainstorm stage means the team can make an informed architectural decision rather than defaulting to the first pattern that comes to mind.

Speaker B: Architectural decisions made early with full option awareness produce far more stable systems. Now a UI concept brainstorm:

Speaker B: Brainstorm ten distinct UI navigation approaches for a complex data analytics dashboard with fifteen modules. For each approach describe the navigation pattern, such as left sidebar, top mega-menu, hub-and-spoke cards, or progressive disclosure, the cognitive load trade-off, and which user type of power user versus occasional visitor it suits best. Include at least two unconventional approaches that challenge standard dashboard conventions.

Speaker A: Requesting at least two unconventional approaches prevents the brainstorm from collapsing into a list of standard patterns. It forces genuine creative range.

Speaker B: Exactly right. You have to explicitly request divergence or the model defaults to convergent safe answers. Now a colour theme brainstorm for a design system:

Speaker B: Brainstorm six distinct colour palette directions for a fintech startup targeting young professional investors. For each palette provide a name, a three-word mood descriptor, the primary and secondary hex colours, a rationale connecting the palette to the target audience's psychology and the brand positioning, and a one-sentence warning about a potential misuse or accessibility risk with that palette.

Speaker A: Including a one-sentence warning per palette is a clever addition. It forces the model to think critically about each option even within an otherwise generative phase.

Speaker B: Bounded criticism at the output stage rather than the generation stage. The brainstorm produces the options; the warnings surface the risks without killing the options. Now a full application concept brainstorm:

Speaker B: Brainstorm five completely different application concepts that could be built around a gym members biometric data feed including heart rate, sleep, recovery score, and workout logs. For each concept describe the core user value proposition, the primary user persona, the monetisation model, the key technical challenge, and the one feature that would make users tell their friends about it. Do not restrict to obvious fitness app concepts.

Speaker A: The explicit instruction not to restrict to obvious fitness app concepts is the key prompt engineering move there. Without it you get five variations of the same dashboard.

Speaker B: Constraint-busting instructions are essential in brainstorm prompts. And to close, a product naming brainstorm:

Speaker B: Brainstorm fifteen name candidates for a B2B project management platform targeting creative agencies. For each name provide the word origin or construction logic, a one-sentence brand personality description, a domain availability assessment noting whether dot-com is likely available, and a one-sentence risk note covering potential trademark conflicts or negative connotations in major English-speaking markets. Include names across a spectrum from abstract and invented to descriptive and literal.

Speaker A: Asking for both abstract invented names and descriptive literal ones across the same brainstorm ensures the naming exploration covers the full creative spectrum rather than clustering around one style.

Speaker B: Brainstorm in a prompt is the single word that tells the model to maximise variety, volume, and creative range. It consistently produces richer, more actionable ideation output than generate or suggest because it activates the model's divergent reasoning mode rather than its convergent selection mode. Master that distinction and your ideation prompts become a genuinely powerful development tool.