Rikyū: The AI Logo Tool That Draws With Geometry, Not Just Pixels

Rikyū, an AI design tool created by Kaisei Suzuki, shipped a logo feature in August 2026 that builds marks from geometric primitives — circles, straight lines and a construction grid — and outputs a downloadable vector with the grid shown and a written concept explanation attached. Rikyū positions the tool as "the AI design tool for businesses without a designer," covering logos, ads, print, websites, apps, 3D and video from a plain description, free to start. The geometry-first approach contrasts with diffusion-based image generators that produce flat PNGs which cannot be scaled cleanly or edited afterward.

Rikyū: The AI Logo Tool That Draws With Geometry, Not Just Pixels Most “AI logo” tools are image generators wearing a costume. You describe a logo, a diffusion model paints something that looks like one, and you get a PNG — flat pixels, no structure, no way to nudge a curve or re-space a letterform without starting over. Rikyū https://rikyu.ai does something different, and it’s worth paying attention to. It Builds With Shapes, Not Pixels Rikyū’s logo feature — shipped in August 2026 by its creator, Kaisei Suzuki — constructs a mark from actual geometric primitives . You give it a description, and it genuinely designs using circles, straight lines, and a construction grid, assembling the logo the way a human brand designer would: diagonals, guide circles, proportional relationships. The final output is a downloadable vector , with the construction grid shown and a written concept explanation attached. That last part matters more than it first appears. One of the hardest problems with generative design isn’t quality — it’s trust . A client looks at an AI-generated mark and has no idea why it’s that shape, so they can’t believe in it. Rikyū’s answer is to make the design process visible : here are the circles, here are the diagonals, here’s how the letter was constructed. The reasoning is the reassurance. Why This Is the Interesting Direction Here’s the argument, and I think it’s right: If AI design ultimately just means generating images faster, the difference from Midjourney or GPT Image is quite limited. The leap comes when AI starts to genuinely understand proportion, geometry, layout, visual systems, and design rules. That’s the line between a faster image maker and a real design tool . Diffusion models operate on appearance — they predict what pixels should look like. A geometry-first system operates on structure — it reasons about relationships that stay valid at any scale and remain editable afterward. The practical payoffs fall out immediately: - Clean scaling — vectors, not raster, so the mark is crisp on a business card and a billboard alike. - Editability — geometry can be adjusted; a flattened PNG can’t. - Explainability — a construction grid and rationale you can show a stakeholder. - Consistency — understanding a visual system is the prerequisite for applying it across ads, print, web, and app surfaces which is Rikyū’s broader pitch . The Bigger Picture Rikyū positions itself as “the AI design tool for businesses without a designer” — logos, ads, print, websites, apps, 3D, and video from a plain description, free to start. The multi-surface ambition is nice, but the geometric logo feature is the part that signals where generative design should head. It’s a small reframing with a large implication: stop asking a model to render a logo, and start asking it to design one — to work in the same structured, rule-governed space that human designers do. That’s the version of AI design that actually competes with the tools, not just the output. Try it at rikyu.ai https://rikyu.ai .