There's so much at stake in next week's US elections. ???? The Russian state has tried hard to influence the outcomes, sowing disbelief and discord. The Zep AI (YC W24) team built a visual tool to explore Russia's recent activities. View the Explorer: ?? https://lnkd.in/g7Mh7FmC The app refers to 50+ sources, including US DOJ indictments, US and foreign government research, non-governmental organization research, and media articles. Offering users a detailed view of these malign influence operations. The Explorer uses Graphiti, Zep's open-source Knowledge Graph library. The AI Assistant was built with LangChain's LangGraph framework and FastHTML. We've spent hours delving into the data. ?? We hope you find it as fascinating as we did. Let us know what you think!
关于我们
Long-Term Memory for AI Assistants. Recall, understand, and extract data from chat histories. Power personalized AI experiences.
- 网站
-
https://www.getzep.com/
Zep AI (YC W24)的外部链接
- 所属行业
- 软件开发
- 规模
- 2-10 人
- 类型
- 私人持股
- 创立
- 2023
Zep AI (YC W24)员工
动态
-
We missed this a few days ago! Thanks for the shout-out, LangChain!
???Building a LangGraph agent with graph memory The following community examples demonstrates building an agent using LangGraph. Graphiti is used to personalize agent responses based on information learned from prior conversations https://lnkd.in/grvxkTtp
-
-
This is a fantastic overview, Eric Vyacheslav! Thanks for the writing about Zep AI (YC W24)! ?? ??
You can now give your AI models long-term memory that actually learns and adapts over time. Zep is a memory layer that allows your agents to store, retrieve, and update information over time using a temporal knowledge graph. This lets your agents to remember key facts, track changes, and reason about evolving data in real-time. Key features: - Instant memory retrieval: Get relevant data from memory in milliseconds without slowing down your LLM. - Built-in temporal reasoning: Zep automatically updates as facts change, so your agents can adapt to new information without manual intervention. - Framework-agnostic: Integrate Zep with Python, TypeScript, or Go—whatever fits your stack. Try it out: https://lnkd.in/gYgcRX3A
-
We now know how to double the IQ of AI agents! One of AI's biggest weaknesses is to keep up with change. Most solutions that try to implement a memory layer for AI agents don't reason well whenever facts change. The same problem happens with RAG applications, which work well until you have to recreate a terabyte of vectors because your documents have changed. Now, you know one of the reasons AI agents seem to get dumber over time. But, a potential game-changer solution is to use a Knowledge Graph to model an agent's world. In a Knowledge Graph, we have a network of connected points, each representing a piece of information. The lines between these points show how they are related. For example, the attached animation shows a point for "Kendra," a point for "Adidas shoes," and a line connecting them to show that Kendra loves the shoes. Let's assume this is a fact our agent understands. A day later, Kendra sends her friend a message: "My Adidas shoes broke, but Pumas are lit!" The agent could use this information to invalidate the previous fact and replace it in the graph with a new point representing "Puma shoes" and a connection stating that Kendra likes them. This was a very simple example, but imagine how powerful it would be to have an agent with access to an always-evolving memory that stays up to date as information changes. That's how Zep works. It's an open-source library you can connect to any agent framework, model, or platform to build your AI application. In a couple of bullet points: 1. Zep synthesizes any messages from users into a Knowledge Graph 2. It allows you to retrieve any relevant facts extremely fast The best way to summarize it is that Zep will immediately double the IQ of your AI agent because they will do something they couldn't before: to reason about facts that change over time. Here is a link to check it out: shortclick.link/xjw3y0 Thanks to the Zep team for sponsoring this post. You can check out their hosted version or use their open-source Community Edition.
-
-
?? ?????? ?? AI agents need more than just conversation memory—they need to understand who they're helping and why. ?? Launching today: We're connecting conversations with business data in Zep Cloud, making AI interactions more personal and relevant to each user. https://hubs.ly/Q02Vlf3T0
-
Knowledge Graphs can't model time, limiting their use as AI agent memory ???. Paul P. shares how Zep AI (YC W24) built Graphiti ??, an open-source temporal Knowledge Graph library. Super interesting article! ?? ?? https://hubs.ly/Q02RRg1r0
-
-
Hey all! We’ve just open-sourced Zep Community Edition, a knowledge graph-based memory layer for AI agents that continuously learns facts from user interactions and changing business data. https://hubs.ly/Q02RfsBD0 Let us know what you think! We’d love your thoughts, feedback, bug reports, and/or contributions!
-
-
LLM data extraction at scale is hard. ?? Learn how the Zep team built Graphiti, our new temporal graph library. ??
Graphiti, our temporal Knowledge Graph library, was super difficult to build. ?? Zep engineer Preston Rasmussen wrote about our challenges, design decisions, and solutions. ??
LLM Data Extraction at Scale
blog.getzep.com
-
Graphiti's "hard" launch yesterday was a wicked success! Super proud of the Zep AI (YC W24) team! Want to learn more? https://hubs.ly/Q02NQ0Jq0 > With Graphiti, AI Agents always have accurate, current data. Graphiti helps you create and query Knowledge Graphs that evolve over time. Knowledge Graphs have been explored extensively for information retrieval. What makes Graphiti unique is its ability to autonomously build a knowledge graph while handling changing relationships and maintaining historical context.
-
-
Build personalized AI assistants with n8n workflows, Zep AI (YC W24) self-improving memory, and our new Zep memory node! Learn more: https://lnkd.in/dX_sdwKR #n8n #ai #llm
-