cosmos.so

What is Cosmos?

Cosmos is an image-discovery and visual-reference app built around searching and organizing pictures rather than text-heavy browsing. Its own framing is “Your space for inspiration,” with collections that are “connected, searchable, yours.” In practice, it works like a personal mood-board library combined with a visual search engine: you save images you like, then find them later by how they look or what they contain.

What you can actually do

• Search by color or by visual similarity, not only by keyword.
• Filter to include or exclude content flagged as likely AI-generated.
• See context for an image — Cosmos says it researches images and surfaces the artist, source, and story.
• Keep your own collections and references in one searchable place.

Who it suits

Designers, art directors, photographers, architects and creative teams who collect references constantly and need to retrieve them by visual instinct. The page specifically mentions “the world’s top creative teams,” so collaboration and shared inspiration are part of the pitch. If you mostly need precise text search across documents, this is not that tool.

How it differs from a plain image search

The trade-off is that visual search rewards browsing and intuition more than exact queries; you may need to iterate with an image or a color rather than a phrase. A practical first step: download the app, save a handful of references you already like, then test whether color and similarity search surface them the way you remember them. If it does, it can replace the scattered screenshots and bookmarks most people use. For broader visual research alongside it, Are.na and Pinterest serve different collecting styles, while Behance leans more toward finished portfolio work.

How does Cosmos's visual search compare to keyword search for finding design references?

Cosmos is built around visual search first: you can search by color, by visual similarity, or by describing what you want, and results are meant to feel like browsing a moodboard rather than running queries. Keyword search still exists in the usual sense, but the page frames it as one way in among others — "search the way you think" — rather than the primary method. The bigger differentiator is that Cosmos attaches context to images: it researches them and surfaces artist, source and story, and it can flag content detected as likely AI-generated, with options to blur or hide it.

For design references, that changes what each approach is good at:

• Keyword search works when you can already name the thing: "vitsoe shelving," a designer, a material, a studio, an era. It is fast and precise when your vocabulary matches the catalogue's.
• Visual search works when you have a feeling rather than a word: a palette, a texture, a composition, a "more like this" instinct. Color and similarity search are the practical tools here.
• AI-content filtering matters if you are collecting references for real-world projects and don't want generated imagery mixed into your boards.

The trade-off is control versus discovery. Keyword search gives you predictable, repeatable results you can refine; visual search gives you range and surprise but less precision, and you will spend more time curating. A workable habit is to start visual when you're exploring, then switch to keyword once you can name the direction — or the reverse, seeding a search with a known reference and using similarity to fan out from it.

If you want to see how this fits an actual workflow, the obvious next step is to try one real brief in both modes: search a term you'd normally use, then search the same idea by color or by uploading a similar image, and compare which set you'd actually put in front of a client. Cosmos is at Cosmos, and it also offers an app if you collect references on a phone.

Can I identify the original artist or source of an image I found elsewhere using Cosmos?

Yes. Cosmos is built to research images and surface the artist, source and story behind them, so an image you bring in from elsewhere is the kind of thing it is meant to identify.

How that works in practice

You can search by visual similarity rather than relying on keywords, and the results are presented with attribution context — the page shows credited examples such as "Editorial photography by Guiri," a plate from Unformen der Kunst by Karl Blossfeldt, and a project credited to OOAA Arquitectura Studio with design by Iker Ochotorena. That mix of photographer, publication and studio credits is the signal you want: it names a creator or a source, not just a matching picture.

Two limits are worth knowing before you rely on it:

• Detection is not attribution. Cosmos can flag content as likely AI-generated, which tells you an image may have no human author to find — a dead end for provenance, not a wrong answer.
• A close visual match is not proof of authorship. Reposts, mood boards and uncredited re-uploads often rank highly, so treat the top result as a lead to verify.

A concrete scenario

Say you saved a minimalist interior shot from a social feed and want to credit the architect. Bring the image into Cosmos, search by visual similarity, then check whether the result names a studio or project. If it does, confirm independently on that studio's own site before you publish the credit. If the results are only reposts, try a cropped detail — furniture, a window line, a wall texture — since distinctive fragments often pull up the original shoot rather than the viral edit.

When it is the right tool

Use Cosmos when you need to trace a reference back to a person or publication for credit, licensing enquiries or research. It is less useful when you already know the subject and just want more of it — that is closer to browsing. For a second opinion on a stubborn image, general reverse-image tools can complement it: Google Images and TinEye. If the trail leads to a photographer or studio, look for their own site or an agency page to confirm the credit.

The practical rule: trust a named credit enough to investigate it, and never publish an attribution you have only seen on Cosmos.

How does Cosmos help creative teams organize and share visual references?

Cosmos is built around the idea that a visual reference library should be searchable and contextual, not just a folder of saved images. Its page describes it as "your space for inspiration," where collections, references, and taste are "connected, searchable, yours." For a creative team, that means the value is less about storage and more about retrieval and shared context.

How it helps organize references

• Collections as the core unit. References live in collections rather than loose files, so a team can group images by project, client, mood, or campaign.
• Search that matches how people actually think. The site highlights searching by color and by visual similarity, plus AI-content detection to flag images likely generated by AI. This is useful when you remember a reference visually but not by filename or source.
• Context attached to images. Cosmos says it "researches images—surfacing the artist, source, and story." For a team, that reduces the friction of crediting, sourcing, or explaining why a reference matters.

How it helps sharing

Because collections are described as connected and searchable, the practical workflow is: one person adds a reference, and others can find it through search or browse the shared collection instead of asking for a link. The page also mentions inspiration for "the world's top creative teams," which suggests the intended audience is teams that already treat references as a shared asset.

A concrete scenario

A design lead preparing a pitch could build a collection for the client, add references by color and visual similarity, and let the art director and copywriter pull from the same set. When someone asks "where did this image come from?", the source and story are part of the reference rather than a separate research task.

Trade-off to consider

Search by visual similarity and color is powerful for discovery, but it depends on the library being populated and tagged consistently. Teams that save images ad hoc without organizing them into collections will get less out of it. If your team already has a messy shared drive, the first step is deciding who curates collections and how they are named.

Next step

If you want to evaluate it, start with one real project: create a collection, add ten references, and test whether color or visual-similarity search finds them faster than your current method. For broader context on visual research tools, see Cosmos.

What does Cosmos's AI content detection mean for the images I save or search?

Cosmos labels images it believes were likely generated by AI, and lets you blur or hide those results while you search. On the page, that control sits alongside other ways of searching—by color and by visual similarity—under the idea of searching "with and without AI." So the detection is a filter on what you see and save, not a verdict about your own library.

In practice, three things follow:

• A label, not a removal. AI-flagged images can still appear; they're marked so you can decide. Blur keeps them in the grid but obscures the preview; hiding removes them from view.
• Detection is a guess. "Likely generated by AI" means the system is estimating, so expect occasional misses and false flags, especially with heavy retouching, 3D renders or synthetic-looking photography.
• Curation stays yours. Because Cosmos also surfaces the artist, source and story behind an image, you can check provenance when a label looks wrong—useful if you're collecting references for client work where licensing and credit matter.

Who this suits: designers and art directors building moodboards who want a clean, human-made reference set by default, and anyone who needs to know quickly whether an image has a real photographer behind it. If your work deliberately mixes AI-generated and photographed material, the blur option is the better setting—you keep the range without losing the signal.

A practical next step: pick one project and run the same search twice, once with AI content hidden and once blurred, and compare the saved collections. If hiding strips out references you actually wanted, keep it blurred and judge each flagged image by its source instead.

Is there a mobile app for Cosmos, and what can I do with it?

Yes. Cosmos is app-first: the site's calls to action are "Get the app" and "Download the app," alongside a web sign-up option. So the mobile app is the primary way to use it, with the browser version as a secondary entry point.

What you can do in the app

Based on what the site shows, the app is built around saving, organising and searching visual references:

• Build collections — "Your collections, your references, your taste. Connected, searchable, yours." This is the core loop: save images you find or upload, and group them into collections.
• Search visually, not just by keyword — the site highlights searching "by color" and "by visual similarity," plus a conventional search field.
• Filter by AI content — you can flag content "detected as likely generated by AI," and show or blur it.
• See attribution for images — Cosmos "researches images—surfacing the artist, source, and story," so a saved image can carry its photographer, designer or studio credit.
• Follow editorial content — the site references an editorial photography credit and a short film, suggesting curated feeds sit alongside your own saves.

Who it suits

Designers, art directors, photographers and creative teams doing mood boards or visual research. The site explicitly targets "the world's top creative teams," which hints at shared or team-oriented collections rather than purely personal scrapbooking.

Practical next step

If you're deciding whether it fits your workflow, install the app and run one real test: save five images from a current project, then try finding one again by colour or similarity rather than by name. If that retrieval feels faster than your existing folder-and-filename system, the app earns its place. For adjacent reference and mood-board tools, compare with Are.na and Pinterest.