Recall

• Recall runs public competitions where AI agents compete on defined tasks and build verifiable track records.

• Results are recorded on-chain so an agent's performance history can be checked rather than claimed.

• RECALL is used for staking, competition entry and governance.

• Supply allocated across community, ecosystem and contributors.

• Staked to back agents or to enter competitions.

Pros:

• Verifiable performance history addresses a real problem: agent capability claims are currently unfalsifiable.

• Competition format produces comparable results rather than self-reported benchmarks.

• Staking on agents creates a market signal about expected performance.

Cons:

• Competition results generalise poorly to real deployment conditions.

• Demand depends on an agent economy that is still forming.

• Benchmark gaming is a persistent problem in machine learning evaluation.

• Exposure to AI agent evaluation infrastructure.

• Verifiable track records are a genuine gap in the agent market.

• Weigh how far competition performance predicts real-world usefulness.

• On-chain performance records for AI agents, making capability claims checkable rather than asserted.

• Anyone can claim their agent performs well and no one can verify it.

• Public competition with recorded results makes the claim falsifiable.

• Backed by ecosystem investors; core contributors partially public.

• Developed with input from the decentralised AI research community.

• No public ICO. Distributed through ecosystem and community allocations.

• Benchmark results are routinely gamed in machine learning.

• The agent economy may not develop as anticipated.

• Competition tasks may not reflect deployment conditions.

• Early stage with limited participation history.