Synorigin

How it works

You paste an idea. Synorigin tells you what it is closest to, how much distinct space it holds, and one concrete thing that would move it.

The four things a result contains

1. Closest matches - up to three

Each match names a specific existing work, product, brand or established convention, with one sentence on the structural or conceptual overlap. Not "this feels similar" - what is actually shared.

Matches are deliberately not all drawn from the most famous possible examples. Search results surface what is most documented, not what is most similar, so at least one match is chosen to sit closer to the scale of your idea where such a match exists - a peer rather than only a giant.

If nothing close genuinely exists, it says so rather than padding the list to three.

2. The originality gap

A plain statement of which element is doing the derivative work. Not a general impression - the specific part.

3. One pivot

A concrete change, precise enough to act on. Not "make it more unique" but the specific thing your version is missing. Where it can be said honestly, the pivot also names adjacent ground that existing work does not cover.

4. A verdict

A short, direct read, including a rough percentage. This stays honest when the honest answer is discouraging - see the chosen-one example, where the verdict came back at roughly 85% a known genre skeleton.

It searches the live web

Every check runs a live web search rather than relying only on what a language model already knows from training. That matters for recent work and for smaller projects that are documented online but were never famous enough to be memorised.

It says so when nothing matches - and that stays rare

A clean result is only worth having if it is hard to earn. When a search genuinely finds no significant precedent, Synorigin states that plainly rather than manufacturing a gap to fill the format. That outcome is held to the same evidentiary standard as reporting a real match: it is said when it is true, not when the check is uncertain or when nothing famous happened to come up.

The two invented-name checks show this working - the invented word was reported as clear, while the generic suffix around it was still called out.

Trademark risk is escalated differently

If a match is an actual registered trademark or active brand in the same category - not merely a shared theme - that is stated as a legal matter, in stronger language than ordinary creative overlap. The Westfield check shows the difference plainly. Ordinary similarity is never dressed up as a legal risk, because a warning that fires constantly stops meaning anything.

It works on more than fiction

The same reasoning handles a character concept, a plot twist, a product idea, a brand name and a marketing angle, because all of them have structure. The recorded examples deliberately span all of these.

What it will not do

• It does not predict success, market timing, or financial outcome. Those cannot be read from a concept.
• It does not evaluate execution quality.
• It does not write the work for you. It identifies the gap; it never builds the plan.
• It does not reproduce copyrighted text - not lyrics, not passages - even when asked directly. It names the work and describes it instead.
• It is not a legal opinion and not a substitute for a trademark or patent search.

The score is an estimate

The originality figure is an AI-generated estimate meant as a starting point for your own judgement, not a measurement. There is no universal scale on which an idea is objectively a given percentage original. Two works can share a chosen one, a mentor and a prophecy and still be completely different stories - which is why the reasoning underneath the number matters more than the number.