PublicAI positions itself as a global, decentralized "human layer of AI": a Web3-enabled human-in-the-loop (HITL) data marketplace and HITL network for training and inference. The site emphasizes a verified contributor network, on-chain staking/validator mechanisms, AI-assisted labeling workflows, and multi-modal data support. PublicAI describes roles and governance including task publishers, workers, reviewers, validators, labeling guilds, approvers, and on-chain councils; tokenomics are centered on a $PUBLIC token with planned token sale activity, staking programs, and buybacks as reported on site. Workflow and tooling described on-site include AI-assisted annotation (AI pre-labels with AI and human validators), model evaluation and scenario testing tools, and a Data Hunter browser extension for social/web capture. Supported data types claimed on the site include image (2D semantic segmentation), text collection, document transcription, named-entity recognition, video, LIDAR/3D, audio, and mapping. Security and privacy statements on the privacy page reference encryption in transit and at rest, access controls, internal policies, and regional privacy rights compliance (CCPA and EEA/UK/Switzerland). Pricing presented publicly is summarized as a free tier plus usage-based (pay-as-you-go) marketplace and an enterprise tier (PublicAI Pro) with no public per-unit or enterprise prices; the business PDF contains marketing statements but no clear per-unit rates. Note: numbers cited on the site (contributors, validators, revenue, etc.) vary across pages — the following aggregates and claims are reported by the site and/or documentation rather than independently verified.