AI Agents and Memory - The presentation focuses on the importance of memory in AI agents, emphasizing that memory is crucial for making agents reflective, interactive, proactive, reactive, and autonomous [6] - The discussion highlights different forms of memory, including short-term, long-term, conversational entity memory, knowledge data store, cache, and working memory [8] - The industry is moving towards AI agents and agentic systems, with a focus on building believable, capable, and reliable agents [1, 21] MongoDB's Role in AI Memory - MongoDB is positioned as a memory provider for agentic systems, offering features needed to turn data into memory and enhance agent capabilities [20, 21, 31] - MongoDB's flexible document data model and retrieval capabilities (graph, vector, text, geospatial query) are highlighted as key advantages for AI memory management [25] - MongoDB acquired Voyage AI to improve AI systems by reducing hallucination through better embedding models and re-rankers [32, 33] - Voyage AI's embedding models and re-rankers will be integrated into MongoDB Atlas to simplify data chunking and retrieval strategies [34] Memory Management and Implementation - Memory management involves generation, storage, retrieval, integration, updating, and forgetting mechanisms [16, 17] - Retrieval Augmented Generation (RAG) is discussed, with MongoDB providing retrieval mechanisms beyond just vector search [18] - The presentation introduces "Memoriz," an open-source library with design patterns for various memory types in AI agents [21, 22, 30] - Different memory types are explored, including persona memory, toolbox memory, conversation memory, workflow memory, episodic memory, long-term memory, and entity memory [23, 25, 26, 27, 29, 30]

I remember lying on the narrow bed of my college town. Eyes fixed on the slow the hypnotical spin of the ceiling fan above me. It was a first year.I hadn't written a single line of working code. I didn't have the clear direction and no one around me had any reason to believe I was building towards something meaningful. And that question kept repeating in my head was what if I go home with nothing. What if I fail publicly and completely.That fear didn't visit occasionally. It moved in. I wasn't just about a ...

Company Overview - Box is a B2B company operating as an unstructured data platform, serving large enterprises including Fortune 500 companies [1][2] - Box has over 115,000 companies as customers, tens of millions of users, and manages over 1 exabyte of data [2] - Box is often the first AI deployed within large enterprises due to existing trust relationships [3] Data Extraction Evolution - Box initially used a straightforward architecture for data extraction involving pre-processing, OCR, and large language models [8] - The initial AI deployment processed 10 million pages, but encountered challenges with complex documents, OCR accuracy, language variations, and the need for confidence scores [9][10][11] - The company experienced a "trough of disillusionment" as the initial AI solution proved insufficient for diverse customer needs [12] Agentic Approach Implementation - Box re-architected its data extraction process using a multi-agent approach, separating problems into sub-agents [12] - The agentic system intelligently groups related fields, dynamically determines data extraction methods, and incorporates a quality feedback loop for continuous improvement [13] - This approach allows for easier updates and specialization, enabling the company to quickly adapt to new document types and customer requirements [13] Engineering and Customer Impact - Building agentic systems helps engineers think about AI and agentic workflows, leading to better understanding of customer needs [13] - This approach facilitates the development of tools that integrate with customer-built agents, enhancing the overall ecosystem [13] - The company advises building agentic systems early when developing intelligent features [14]