A daily crypto, finance, and AI vocabulary grouping puzzle for beginners.
RWA Reporting Rails
These concepts describe operational records around tokenized funds and cash-like instruments without recommending any product.
Fund LookthroughIssuer RegistryRecord DateReserve Reconciliation
- Fund Lookthrough: Fund lookthrough is the process of examining the underlying assets or exposures inside a fund rather than only viewing the fund wrapper.
- Issuer Registry: An issuer registry is a structured record of entities authorized or recognized to issue a tokenized asset, security, or financial instrument.
- Record Date: A record date is the cutoff date used to determine which holders are recorded for an event such as a distribution, vote, or report.
- Reserve Reconciliation: Reserve reconciliation compares issued token balances with backing asset records to identify mismatches, timing gaps, or reporting issues.
DeFi Liquidation Risk
These terms explain risk controls and automated actions used when collateralized loans become under-secured.
Oracle LatencyLiquidation QueueCollateral HaircutKeeper Bot - Oracle Latency: Oracle latency is the delay between a market event and the time a protocol receives or applies the updated external data.
- Liquidation Queue: A liquidation queue is the ordering or process a protocol uses to handle positions that fall below required collateral thresholds.
- Collateral Haircut: A collateral haircut is a discount applied to an asset's market value when calculating how much borrowing power it should support.
- Keeper Bot: A keeper bot monitors protocol conditions and submits transactions for routine tasks such as liquidations, rebalances, or updates.
Agent Memory Systems
These ideas describe how agent applications keep useful context available while controlling what is retrieved or summarized.
Conversation MemoryVector StoreRetrieval PolicyMemory Summarization
- Conversation Memory: Conversation memory is saved context from previous interactions that an AI system may use to maintain continuity in later responses.
- Vector Store: A vector store is a database optimized for saving and searching embeddings so applications can retrieve semantically related information.
- Retrieval Policy: A retrieval policy defines what information an AI application should search for, rank, include, or exclude before generating an answer.
- Memory Summarization: Memory summarization condenses older conversation or task history into shorter notes that preserve useful context for future model calls.
AI Data Marketplaces
These concepts describe how datasets may be packaged, licensed, tracked, or permissioned for model development and evaluation.
- Data Licensing: Data licensing defines the rights, limits, and obligations attached to using a dataset for analysis, training, evaluation, or products.
- Synthetic Data: Synthetic data is artificially generated data designed to resemble useful patterns from real data while serving a specific testing or training purpose.
- Dataset Lineage: Dataset lineage tracks where a dataset came from, how it was transformed, and which versions or sources contributed to it.
- Consent Receipt: A consent receipt is a record showing that permission was granted for specific data use under defined terms or conditions.