💰HAI Tokenomics

The HAI token is the core of the incentive model, powering the reinforcement learning engine and bringing high quality human feedback. Designed after inspiration from DePIN projects, HAI will be distributed to platform contributors fairly and gradually over time. HAI token holders are the net beneficiaries of the token value accrual mechanism - buyback token pool and ETH yield distribution (to be implemented later this year).

A couple of challenges that we need to tackle early is spam prevention and attaching reputation to feedback providers. The pseudonymous nature of contributors requires us to provide reputation based on the work completed by contributors to their wallet address. HAI token staking is similar to a proof of stake mechanism wherein users are ready to lock up tokens in order to honestly contribute to the reinforcement learning process and token gating also work as a DOS attack mitigation tool.

The HAI token has a capped supply of 1,000,000,000 (1 billion) which becomes deflationary when protocol revenue is higher than incentive spends. The HAI token has multiple roles to play in this economy: staking mechanism, reward system and governance

Network partcipant	% of token supply	Unlock schedule
Team	12%	Cliff - 1 year, followed by 2 years linear unlocks
Strategic Partners	12%	Cliff - 6 months, followed by 1 year linear unlocks
Airdrop	6%	Unlocked at TGE
Liquidity Provision	10%	50% unlocked at TGE, followed by 6 months linear unlock
Incentive programs	50%	Cliff - 3 months, followed by 3 years linear unlocks
LBP sale	10%	Unlocked at TGE

Incentives programs are just reward distributions to contributors for providing feedback on topics where they believe the LLM lacks capabilities and can suggest improvements to train the models further. Initially these rewards will be distributed on a bi-weekly basis in HAI tokens. We are in discussions and expect LLM projects to also reward users in their own tokens or via stablecoin/ETH payments in the future.