All posts

May 23, 2026

How to Get Multiple Design Directions Before Committing to One

The right way to evaluate a design isn't to iterate from a single output — it's to explore several distinct directions at once. Here's how to generate five different takes on the same product, and why the comparison is where the real decision happens.

The standard AI design workflow goes like this: describe what you want, get a generation, iterate toward something better. The premise is that you start from one output and refine it. The problem with this premise is that refining one direction doesn't tell you whether it was the right direction to start with.

Getting five different takes on the same product isn't about having more options to agonise over. It's about creating the conditions under which good design decisions can actually be made. You can't evaluate a direction until you can see it alongside alternatives. You can't know whether "clean and minimal" is right for your product until you've also seen what "warm and editorial" looks like on the same screens, with the same content, in the same information hierarchy.

Why single-path iteration produces shallow decisions

When a design process starts with one generated output and iterates from there, the aesthetic decisions are made implicitly. The first generation came with a colour palette, a type system, a spacing approach, a visual language. Maybe they're good. Maybe they're just what the AI defaults to. Either way, from that point on, every subsequent iteration is a variation on those decisions rather than a reconsideration of them.

The diverge-then-converge principle is foundational in design methodology: you need range before you need refinement. Skipping the range step doesn't make the process faster — it makes every subsequent decision shallower because you're optimising a direction you never fully tested against alternatives. The designer who starts with three or four distinct concepts and selects the most promising is working from a stronger position than the one who started with one and improved it. The improvement of a wrong direction is still a wrong direction.

This matters practically when you're presenting to stakeholders or clients. A conversation about which of three distinct design directions best serves the product is a different quality of conversation from a conversation about whether to make the header taller. The first produces alignment on what the product should feel like. The second produces minor adjustments. If you arrive at a client presentation with a single design, you're asking them to approve or reject a fait accompli. If you arrive with three, you're inviting them into a decision.

What "different directions" actually means

Five different takes on the same product doesn't mean five colour palette variations. That's not direction exploration — it's palette swapping. Real direction exploration means different visual languages: a direction that leans into data density vs. one that prioritises whitespace; a direction with an editorial, typographically-driven feel vs. one that uses illustration or strong iconography; a direction that feels like a professional B2B tool vs. one that feels warmer and more consumer-leaning.

The differences that matter are the ones that reflect different bets about how the product should feel to its users. Different bets can be right or wrong for a given product and audience. The way to figure out which is right is to see them both, applied to the same real product, at enough fidelity that your aesthetic judgment can actually engage.

Abstract palette cards don't give you that fidelity. You need to see the direction on your actual screens — your real content structure, your information hierarchy, the way the data you're presenting actually looks inside the UI — before you can know whether the direction works for this product or just works in the abstract.

How the moodboard generates and compares directions

After the product questionnaire, before a single screen is generated, Mowgli's moodboard surfaces 16+ distinct design directions tailored to your specific product. These aren't generic templates. Each direction is a fully realised visual language — typography, colour, spacing, tone, naming — applied to content drawn from your actual product spec.

You browse the full grid. Browse without commitment: the cost of looking at a direction and deciding it's not right is zero. Select up to four directions that catch your eye and they stack in a shortlist panel. Then the preview step: each selected direction is applied to a flagship screen from your actual app — your real layout, your real content, your real information hierarchy — so you can see what it actually looks like as a live product, not as an abstract concept.

This is the comparison step. Four directions, side by side, on the same screen. You're not imagining what they'd look like. You're looking at them. You go back to the moodboard, swap one out, try a fifth. You upload a reference — a screenshot of something with a feeling you're trying to capture, a brand you admire, a photo that suggests a tone — and the moodboard generates directions that respond to it. The loop is fast enough that you're genuinely exploring, not just approving.

For client presentations, this process gives you something concrete: two or three genuinely distinct directions for the same product, each applied to real screens, each ready to show. The conversation becomes "which of these is right for what you're building?" — which is the right question to be answering.

On the canvas: ask for variations any time

The moodboard is the right tool for exploring directions before your product is generated. But exploration doesn't stop there — and if you're already on the canvas with a generated product and want to try something different, you don't need to go back to the start.

Mowgli's canvas chat is a product and design-aware co-pilot specialised on your specific product, on top of your design file. Ask for what you want and it renders directly on the canvas as an alternative screen, next to what you already have. "Give me five variations of this screen with different visual directions." Five screens appear on the canvas. You zoom in, compare layouts, inspect the details, see them all at the same time. When one catches your eye, "apply this direction to all screens" and the whole product updates.

This makes exploration available at any point in the process — not just before generation. A new idea mid-project, a client who wants to see an alternative direction, a stakeholder who asks "what would this look like if it felt more minimal" — none of these require starting over. The canvas is always open for exploration. The comparison is always spatial, always immediate.

After the direction is chosen

Once you've compared directions and selected one, the full generation happens from that point. The aesthetic decision is already made — deliberately, from options you actually evaluated, with your own eye — before any of the detailed design work begins.

The full build then produces every screen the product needs: 20, 30, 40+ screens, all in the direction you chose, all derived from the same product spec. The coherence comes from the spec: every screen expresses the same product model, in the same visual language, because both were established before generation began.

If you're presenting to a client and want to leave the aesthetic decision to them, you can run the moodboard in two directions, generate two full product builds, and arrive with two complete, screenworthy design sets. The comparison is genuine — same product, same spec, two distinct visual languages — and the client is choosing between real things rather than hypotheticals.

Sources