Tag
This paper introduces ISEE, an interactive system that uses LLM-based agents to assess and collaboratively enrich the semantic quality of database field descriptions, reducing cognitive load and improving downstream tasks like entity linking.
A discussion on whether LLM capability or enterprise data access and permissions is the bigger bottleneck for AI adoption, arguing that content management is becoming as critical as the models themselves.
Airtable and OpenAI have partnered to integrate Airtable directly into ChatGPT, enabling users to connect databases, generate tables, and manage data within the chat interface.
This paper proposes ASMR, an agentic framework with a Field Generation Agent and a Structural Optimizer Agent that uses reinforcement learning to automatically generate compact and informative schemas from historical ship maintenance reports, aiming to improve report completeness and consistency.
This paper introduces Agentic Data Environments, focusing on the shift from read-only agents to read-write agents that mutate environments with consequences. It discusses the value proposition of agentic automation and the asymmetry between benefits and catastrophic costs of failures.
Compendium is a new product that helps keep teams, agents, and data organized on one page, launched on Product Hunt.
Zoho Tables offers a smarter way to manage work and data.
This paper from SJTU and Tsinghua systematically evaluates 12 agent memory systems from a data management perspective, decomposing memory into four modules and providing guidelines on when to use RAG, vector databases, or knowledge graphs for long-term agent memory.
A paper systematically evaluates 12 LLM Agent memory systems, breaks them into four modules, finds no single architecture dominates all scenarios, and reveals cost-performance trade-offs and common issues (e.g., 'past hallucinations').
This paper presents a systematic experimental study of agent memory systems from a data management perspective, decomposing memory into four core modules and evaluating 12 representative systems across 11 datasets, finding no single architecture dominates and highlighting cost-performance trade-offs.
Merlin by Encord enables users to manage AI data infrastructure through a single conversation interface.
A research paper proposing a unified agentic-retrieval framework for autonomous context-aware data quality assessment. It interprets natural-language usage descriptions, generates executable validation logic via multi-agent workflow, and uses feasibility validation to ensure reliability.
This paper describes Scuba, a distributed in-memory database system developed at Facebook for real-time analytics and data exploration.
A reflection on agent memory as primarily an infrastructure/data-management problem rather than an AI problem, focusing on practical complexities like permissions, scopes, and revision history.
This paper rethinks the data foundations for long-term AI agent memory, arguing that current database paradigms fall short. It introduces Governed Evolving Memory (GEM), a formalization with state-level operators and correctness conditions, and presents a prototype called MemState built on a property graph backend.
A paper argues that agent memory should not be treated as a database with CRUD operations, proposing Governed Evolving Memory (GEM) with state-level operators and a topic graph data model to address issues of integration, propagation, relevance, and adaptation.
This video provides an in-depth introduction to the file organization of IBM System/360, covering sequential and random data processing, indexing mechanisms, device independence, and cost advantages. It is a historical technical review.
Verus is an open-source belief database for AI agents that tracks conflicting claims from multiple sources with confidence scores and conflict detection; the author seeks real-world datasets and feedback.
Claro is a product that enables AI agents to operate on user data, positioning itself as a platform for autonomous data operations.
OpenAI introduces new data management features for ChatGPT, including the ability to disable chat history to prevent model training, a new ChatGPT Business subscription with stricter data policies, and an export tool for users to access their stored data.