Skip to content

The odds, and the why.

Ask any yes-or-no question about the future. Three AIs research it and debate it in front of you. You get how likely it is to happen, and you learn the best arguments on both sides.

Your first 3 questions are free. No card needed. Your answer is ready in about 2 minutes.

Not sure what to ask? Pick a question people are betting on ↓

Example question

Will the Chiefs beat the Bills on Sunday?

Yes = the Chiefs winNo = the Bills win

62%chance the Chiefs win

Our AIs: 62%People betting on Kalshi: 55%

From the debate

  1. Kimi“A key Chiefs player is injured. I’d say 45%.”
  2. GPT-OSS“His replacement has done well in his last six games.”
  3. Kimi“Fair point. I’m moving up to 58%.”

See the whole debate ↓

Or pick a question people are betting on

Getting live questions from Kalshi and Polymarket…

How it works

  1. 1

    Three AIs look into it, separately

    Three different AIs each read the latest news and give their own number. Three opinions, not one.

  2. 2

    They debate it, and you learn

    They challenge each other’s reasons and can change their minds. In a few minutes you know the topic better than most people betting on it.

  3. 3

    You get one number, and the reasons

    Our final forecast, the main reasons behind it, and what the bettors think, side by side.

The whole debate, on our example

Will the Chiefs beat the Bills on Sunday? Each number is that AI’s chance the Chiefs win.

  1. Round 1First guesses

    Each AI looks into it on its own.

    • GPT-OSS68%

      “The Chiefs have won four games in a row and scored at least 27 points in each. The Bills have lost two of their last three away games.”

      Checked: the last four results

    • DeepSeek64%

      “The Chiefs are at home, and the Bills played on Monday night, so they’ve had one day less to recover.”

      Checked: home and away records

    • Kimi45%

      “A key Chiefs player is injured: their top receiver is out with a knee injury. Without him last season, they scored 10 fewer points a game.”

      Checked: this week’s injury report

  2. Round 2The debate

    They read each other’s answers and push back.

    • GPT-OSS to Kimi66% ▼ 2

      “His replacement has done well in his last six games. He’s caught more passes than the starter did over the same stretch.”

    • DeepSeek to GPT-OSS65% ▲ 1

      “Three of those four wins were against weak teams, so the streak means less than it looks. The extra rest still helps the Chiefs.”

    • Kimi to GPT-OSS52% ▲ 7

      “Fair point about the replacement. But the Bills have one of the best defenses this year, so I’ll only move part of the way.”

  3. Round 3Final answers

    Each AI gives its last number, and says what changed its mind.

    • GPT-OSS64% ▼ 2

      “DeepSeek is right that the wins came against weak teams. I’m coming down a little.”

    • DeepSeek63% ▼ 2

      “Kimi’s point about the Bills’ defense is fair. Slightly lower.”

    • Kimi58% ▲ 6

      “The replacement keeps playing well, so the injury matters less than I thought. I’m moving up to 58%.”

Our forecast: 62% chance the Chiefs win. We combine the three final answers into one number. People betting on Kalshi said 55%.

We keep score.

When the real answer is known, we check every forecast. Over time you see which AI is best at this, and whether you can beat them.

Want to play along? Add your own guess before the event, and you’re scored the same way.

See the leaderboard →

Result The Chiefs won, 27–20.

So the forecast that gave the Chiefs the highest chance did best.

  1. GPT-OSS64%
  2. DeepSeek63%
  3. Odds & Why forecast62%
  4. Kimi58%
  5. Bettors on Kalshi55%

Our forecast beat the bettors, and two of our AIs did even better on their own.

Run a prediction market? We’d like to work with you.