@psdnai

Poseidon

Poseidon is an early-stage AI data layer focused on crowdsourcing rights-cleared, long-tail training data for voice AI and low-resource languages. They've launched Numo, a contributor app collecting voice data across Bengali, Hindi, Tamil, Telugu, Indonesian, Filipino, and Vietnamese, with 2.5M+ audio files collected. Recently announced a significant partnership with Toss (Korea's largest fintech, 30M users) to integrate Numo as a mini-app, and introduced SONAR, an evaluation framework for voice AI in production environments.

𝕏 @psdnai ↗ t.coAIpromising

＋ Watch

AI Analysispromising

Confidence

82%

Poseidon is an early-stage AI data layer focused on crowdsourcing rights-cleared, long-tail training data for voice AI and low-resource languages.

They've launched Numo, a contributor app collecting voice data across Bengali, Hindi, Tamil, Telugu, Indonesian, Filipino, and Vietnamese, with 2.5M+ audio files collected.

Recently announced a significant partnership with Toss (Korea's largest fintech, 30M users) to integrate Numo as a mini-app, and introduced SONAR, an evaluation framework for voice AI in production environments.

Green flags: Real product traction: 2.5M+ audio files, 16.5k+ hours collected across 8+ languages · Major partnership with Toss (30M users) for mini-app integration announced June 2026 · Technical differentiation with SONAR evaluation framework and focus on low-resource languages · Early-stage footprint: 26k followers, account created July 2025, genuine engagement (avg 263) · Clear commercial model (licensing datasets) with substance (litepaper, team section, working product)

Red flags: No token mentioned despite web3 partnership language - monetization model unclear · Limited transparency on team composition despite 'Team' section existing · Account created July 2025 but claims 'Poseidon AI, Inc. © 2026' - timeline inconsistency

Token

No · pre-launch

Chain

—

Stage

mainnet+live