Create a clash. Get AI verdicts. Let the community vote. See the truth.
Ya-Hi is a platform for structured debates between any two things — products, people, ideas, opinions, or even yourself. Every clash goes through a four-step pipeline that ends with a transparent, side-by-side comparison of AI analysis and human voting. Here is exactly how it works.
Creating a clash takes under a minute. You need three things: Entity A (the first side), Entity B (the second side), and a Title — the question your clash is asking. For example: Entity A = "Notion", Entity B = "Obsidian", Title = "Which is the better note-taking app for developers?"
You can optionally add a hint (a sentence giving context), a category, and images for each entity. The more context you provide, the more specific and accurate the AI verdicts will be.
Within a few minutes of creation, Ya-Hi's AI pipeline kicks in automatically. First, the AI identifies 4 comparison attributes — the key dimensions that matter for your specific clash. For a note-taking app clash, these might be: Collaboration Features, Learning Curve, Mobile Experience, and Plugin Ecosystem.
Then two independent AI engines — Claude (Anthropic) and DeepSeek — each score both entities on every attribute. Scores are from 0–100 and must sum to 100 per attribute (so if Claude gives Notion 70 on Collaboration, Obsidian gets 30). Each engine also writes a reasoning paragraph. You get two completely independent perspectives, not one averaged result.
Once AI verdicts are ready, any logged-in user can vote on the clash. Voting is not a single choice — you vote on each attribute separately. This granularity matters: you might think Notion wins on Collaboration but Obsidian wins on Learning Curve. Your nuanced opinion is captured, not flattened.
Votes are anonymous. Users cannot see the AI scores when they vote, so human opinion is not anchored to what the AI said. This keeps the data honest.
The clash page shows the full picture: Claude's score, DeepSeek's score, and the human vote percentage — for every attribute. A coloured bar makes the split instantly readable. You can see at a glance whether AI and humans agree or diverge, and by how much.
A controversy score highlights clashes where human and AI opinion are furthest apart. These are often the most interesting debates — the ones where data and gut feeling point in opposite directions.
Beyond the classic A vs B format, Ya-Hi supports three other clash types for different situations.
State a bold opinion. AI and community vote True or Cringe. Perfect for controversial takes and unpopular opinions.
Paste a news article URL. AI reads it and generates a verdict grounded in the article's actual content. Great for current events.
Put yourself as Entity A. Choose Fair analysis, Advocate mode (AI argues for you), or Roast mode (AI humourously tears you apart).
Any clash type can be toggled to Fun Mode. The AI uses more playful, meme-inspired language while still providing real scores.
Ya-Hi pays creators for the clashes they build. When your clash earns advertising revenue through views and engagement, 60% of that revenue goes to you. Earnings accumulate in your wallet and can be withdrawn via UPI at any time.
There is no minimum follower count, no approval process, and no cost to create. Any user can earn from day one. The best-performing clashes are ones that tap into a genuine ongoing debate — something people already argue about and want to settle.
No. Anyone can browse the feed and read clash verdicts without an account. You need an account to create clashes, vote, and earn revenue. Signing up takes about 30 seconds via Google or email.
Usually between 2 and 5 minutes after you create a clash. The AI pipeline runs two engines in parallel. You will see a progress indicator on the clash page as each engine completes. Complex topics with more context take slightly longer.
The title and entities are locked once a clash is published, to keep the AI verdicts and human votes consistent. You can update the description and category. If you made a significant error, contact us and we can help.
Ya-Hi shows you the scores — it does not pick a single winner. Each attribute has its own score, and you can see the overall average. The philosophy is that real debates rarely have a clean winner: context matters, and different people weigh attributes differently.
Each AI model has different training data, reasoning patterns, and knowledge cutoffs. Disagreements between the two engines are genuinely informative — they surface the attributes where even well-informed analysis is uncertain. We consider this a feature, not a bug.
Revenue comes from display advertising on the platform. Each month, total ad revenue is pooled and distributed to creators proportionally based on how many views and votes their clashes received that month. The creator receives 60% of the revenue attributed to their content. Detailed earnings are visible in your profile wallet.