Tenki

Tenki is a prediction market AI research tool from Whimsey Labs for Kalshi and Polymarket: it scans thousands of markets across both platforms and flags opportunities where prices diverge from the evidence. Note: the old name AI Sports Picks for Kalshi Traders (NBA/NFL/NHL/NCAA) is outdated — the product has pivoted from pure sports picks to all-category research, and is currently in a Waitlist pre-launch phase, not yet live.

Core capabilities

• Continuous market scanning: monitors thousands of markets on both platforms 24 hours a day, narrowing down to the few opportunities worth a close look
• AI mispricing detection: compares market traded prices against evidence-backed fair probabilities, surfacing divergences for review
• 3–7 daily picks: a short list each morning with evidence, price gaps, and what would have to happen for the call to be wrong
• Verifiable reasoning trails: We show our work — every pick carries bull/bear arguments on both sides plus sources, rejecting black boxes
• Public track record pledge: every settled result will be published, losses included, bad days not deleted (fulfilled only after launch)
• One feed for two markets: side-by-side comparison across both platforms, covering the categories its models handle best (macro/Fed, for example)

Use cases Active traders who live on Kalshi/Polymarket but hold full-time jobs with no time to sweep the board, and data-driven regulars who want a second pair of eyes plus checkable homework; curious newcomers can learn from its mispricing logic. Not for high-frequency trading or anyone needing automatic order placement — it does research only, no orders, no API.

Unique strengths It bills itself as the Bloomberg of prediction markets: it runs no markets and places no trades, only narrow-and-deep research picks. Differentiation lies in verifiable evidence chains + time-aware LLM reasoning, advocating one source for both platforms, with a publish-losses-included transparency pledge ahead of most signal services.

Editorial verdict The concept and founding team (Michael Ulin, with a McKinsey/ZestyAI/Paxton background; related companies have raised $78M+ combined) are credible, but the product cannot be purchased yet and neither track record nor pricing can be verified, with the historical messaging having flip-flopped repeatedly. Overall 2.5/5: register for the Waitlist first, and evaluate paid tiers only after launch and publication of dated results.