Athena
An open-source AI agent memory service that employs a three-tier memory architecture comprising a short-term context window, an intermediate layer, and a long-term knowledge graph. It combines topic segmentation, the Ebbinghaus forgetting curve, and spaced repetition mechanisms to enable dynamic management and active forgetting of agent memory.
Timeline (last 90 days)
Glue is a serverless data integration service responsible for crawling data sources, maintaining the data catalog, and running ETL jobs; Athena is an interactive query engine that runs standard SQL directly against data on S3 based on that catalog and is billed by the amount of data scanned.
The toolkit includes Glue and Athena plugins, letting agents directly handle data query and analysis tasks.
QuickSight's core competitive advantage lies in its deep integration with the AWS ecosystem (such as direct connections to Redshift, S3, and Athena) and its pay-per-use pricing model.
Athena's intermediate layer summarizes concluded old topics and embeds them all at once after the conversation shifts to a new topic.
Athena's long-term graph records recurring people, projects, and tools, forming a continuously evolving relationship network.
Athena rejects the use of TTL (Time-To-Live) periodic cleanup and the "keep the most recent N items" window mechanism.
Athena segments conversations into multiple "chains" by topic by computing the cosine similarity between adjacent conversation turns, closing a chain when the similarity drops.
Athena assigns each closed chain a heat score based on the Ebbinghaus forgetting curve, with the formula Heat = I_base * exp(-ΔT / (τ * S)).
Athena's recall strength S sits in the denominator, increasing by one point each time it is retrieved; frequent use flattens the forgetting curve, while prolonged disuse makes the curve steeper.
Only chains whose heat remains sufficiently high are extracted into Athena's long-term graph.
7 more timeline events
All Facts (17)
The S3 API has given rise to a large toolchain ecosystem, including the AWS CLI, SDKs for various languages, s3cmd, rclone, and others.
60%UnverifiedGlue is a serverless data integration service responsible for crawling data sources, maintaining the data catalog, and running ETL jobs; Athena is an interactive query engine that runs standard SQL directly against data on S3 based on that catalog and is billed by the amount of data scanned.
50%UnverifiedThe toolkit includes Glue and Athena plugins, letting agents directly handle data query and analysis tasks.
50%UnverifiedQuickSight's core competitive advantage lies in its deep integration with the AWS ecosystem (such as direct connections to Redshift, S3, and Athena) and its pay-per-use pricing model.
50%UnverifiedAthena's long-term graph records recurring people, projects, and tools, forming a continuously evolving relationship network.
50%UnverifiedAthena rejects the use of TTL (Time-To-Live) periodic cleanup and the "keep the most recent N items" window mechanism.
50%UnverifiedAthena segments conversations into multiple "chains" by topic by computing the cosine similarity between adjacent conversation turns, closing a chain when the similarity drops.
50%UnverifiedAthena assigns each closed chain a heat score based on the Ebbinghaus forgetting curve, with the formula Heat = I_base * exp(-ΔT / (τ * S)).
50%UnverifiedIn Athena's heat formula, the decay constant τ defaults to 24 hours.
50%UnverifiedAthena's recall strength S sits in the denominator, increasing by one point each time it is retrieved; frequent use flattens the forgetting curve, while prolonged disuse makes the curve steeper.
50%UnverifiedOnly chains whose heat remains sufficiently high are extracted into Athena's long-term graph.
50%UnverifiedIndiscriminate memory accumulation is not only costly but also causes retrieval to be overwhelmed by irrelevant noise, whereas a system that knows how to forget may produce more relevant outputs.
50%UnverifiedThe content related to Athena comes from a single developer's post on Reddit, and its actual effectiveness and applicable boundaries remain to be validated by further practice.
50%UnverifiedThe Athena author has opened a Discord community and holds weekly office hours to discuss the design.
50%UnverifiedAthena is an open-source AI agent memory service that adopts a three-tier memory structure (short-term window, intermediate layer, long-term graph).
50%UnverifiedAthena's intermediate layer summarizes concluded old topics and embeds them all at once after the conversation shifts to a new topic.
50%UnverifiedAthena's short-term window stores recently occurred events, functioning as working memory.
50%