Have you ever had your agent working for almost one hour only to understand that he went in the wrong direction or in the middle of something very important he ran out of context window. Me too. That's why in the last months I developed a workflow that consists in dividing a big feature into smaller markdown tasks.Hi, I'm Alex Gavesco and I'm going to present backlog MD, a tool for project management for AI agents and humans. Okay, let's start. So, have you ever seen a terminal cand board.Well, when I start ...

Right. Hello everyone. Uh today I will present meta adaptive context engineering or meta AC for short which is a new framework designed to optimize AI agents beyond single dimension approaches.We will explore how orchestrating multiple adaptation strategies can overcome the limitations of existing context engineering methods. Now a little introduction about myself. Uh so I'm Alberto Romero.I'm the co-founder and CEO at jointly. And for context at jointly we build the main specialized agents for regulated in ...

AI Model Deployment Challenges - AI 工程团队面临着基础设施复杂性的问题，需要在不同平台和模型之间进行部署，并希望简化流程，使用户能够使用统一的客户端访问任何模型，而无需复杂的代码更改 [2][3][4] - 行业需要一种简单且标准化的方法，使开发人员能够轻松地将其内部构建的 AI 模型或在 GitHub 上找到的开源模型集成到其代码库中，并易于执行 [7] - 行业预测 AI 部署的未来是混合推理，即小型模型在本地或边缘位置与大型云 AI 模型协同工作，因此开发人员需要转向更低级别、更接近硬件且响应更快的解决方案 [8][9] Python Compiler Solution - 该方案构建了一个 Python 编译器，允许开发人员编写简单的 Python 代码，并将其转换为可在任何地方运行的微型自包含二进制文件，包括云、Apple 芯片等 [5] - 该编译器使用 LLM 在编译管道中生成 C++ 和 Rust 代码，从而能够运行各种 AI 模型，并扩展到服务器端以外的更多位置 [6][33] - 编译器通过 tracing 技术生成函数内部所有操作的图表示，最初尝试使用 PyTorch 的 Torch FX，但由于其对 PyTorch 代码的关注和对 fake 输入的依赖而放弃，转而使用 LLM 生成 traces，最终通过分析 Python 代码的抽象语法树并使用内部启发式方法构建内部表示 [13][14][15][16][17][18] - 编译器采用类型传播技术，通过分析 Python 函数的签名和 C++ 的原生类型信息，推断并约束生成代码中的变量类型，从而解决 Python 动态类型与 C++ 静态类型之间的差异 [25][26][27][28] Implementation and Usage - 通过类型信息传播，编译器能够生成正确的 C++ 代码，并将其编译为可在任何设备或平台上本地运行的动态库 [34][35][36] - 可以使用 FFI（外部函数接口）从 JavaScript 和 Node.js 调用编译后的库，从而允许在各种环境中使用编译后的 AI 模型 [37][38][39] - 通过创建一个 OpenAI 风格的客户端，可以将编译后的嵌入模型暴露给用户，从而使用户能够像使用官方 OpenAI 客户端一样访问任何开源模型 [40][41]

Hi, I'm Offer from Victara. At Victara, we developed a trustworthy agent operating system. And there's a lot of really cool use cases with this like document generation, conversational AI or chat bots, either internal or external, and enterprise deep research, which I'm going to talk about today.But before I jump into Enterprise Deep Research, let me tell you a little bit more about our operating system for agents. First of all, it's a SAS platform, but also runs on your own VPC or on premise in your own da ...

Pantic AI Products & Features - Pantic AI supports temporal and other durable execution frameworks, with ongoing efforts to integrate more workflow orchestration backends [1] - Pantic AI offers tools for building AI agents, including the ability to perform web searches and analyze data [11][41] - Pantic AI's temporal agent handles the IO needed to call an LLM, including tool calls, by turning them into activities [16] - Pantic AI is developing a gateway for buying inference from various models, including observability features [61] Temporal & Durable Execution - Temporal is highlighted as a leading solution for durable execution, crucial for long-running workflows where progress preservation is essential [2] - Temporal records every activity and its inputs/outputs, enabling rerun from any point by plugging in the answers [15] - Temporal enables the resumption of workflows without adding resume code to the agent code [29] - Temporal's retry logic handles runtime errors and ensures continuous operation [22][25] Deep Research & Agent Architecture - Deep research is presented as analogous to a 20 questions game, with web search or RAG as intermediate steps [11] - The company is shifting towards viewing agents as micro-tasks that form larger autonomous task completion systems [40] - A deep research agent can be composed of multiple specialized agents, such as a plan agent, a search agent, and an analysis agent [41] Evaluation & Performance - Pantic AI evals are used to compare the performance of different models, considering factors like cost, speed, and accuracy [33] - Gemini was initially found to be faster and cheaper, but later discovered to sometimes invent incorrect answers [33][35]

This is Valberkichi, Weta's chief AI officer, and I am joined by >> Kellen Fox, head out of the product management team here at WA >> and we're both thrilled to present context platform engineering to you at the AI. engineering code summit. Now, let's kick this off with uh an announcement we're making.We're actually open sourcing our context platform engineering toolkit. And this toolkit features a really cool load generator that Kalen wrote that lets you configure agent swarms uh and agent subtasks with ve ...

Hi everybody, my name is Brian John and I'm excited today to talk to you about hacking sub agents in the codeex CLI. So who am I. I'm a principal fullstack engineer.My current focus at work is AI enablement for R&D. So think helping our R&D team members get their work done faster and with higher quality using AI. The company I work for is BetterUp.It's an awesome place to work. We've been using AI since the very beginning. I've been there for over eight years now, which is longer than any place I've ever wo ...

Product Launch & Core Concept - Command Code, a coding agent with taste, is launched after over a year of development [1] - The core idea is to build a coding agent that learns and adapts to a programmer's coding style and preferences, creating a personalized coding experience [3][4] - Command Code aims to address the sloppiness and lack of personalization in existing LLM-based coding tools [12][14][16] - Taste models are introduced as a way to capture and share coding intuition and intentions, potentially revolutionizing code generation [28] Technology & Architecture - Command Code utilizes a meta-neurosymbolic architecture with reinforcement learning to create a deterministic and explainable system [9][22][23] - The architecture combines LLMs with a "taste" model, which is a representation of the user's coding preferences [24] - Reflective context engineering enables the system to continuously learn and adapt to the user's evolving coding habits [25] - The system uses both explicit and implicit feedback to refine the neurosymbolic space and enforce the user's coding choices [24] Business & Market Opportunity - Langbase, the company behind Command Code, raised $5 million from investors [11] - The company aims to build a large ecosystem around taste models, enabling developers to share and leverage coding styles [27] - Command Code has shown internal gains at Langbase, with a 10x increase in code merged to the main repository and a significant reduction in code review time [32] - The product targets individual developers, teams, and enterprises seeking to improve coding speed, consistency, and maintainability [29]

Context Engineering & AI - Context engineering is evolving from simple prompt engineering to a dynamic approach that feeds AI with wider context for better results [3] - Context engineering enables selective curation of information relevant to specific domains, especially important in enterprise environments [4] - Structuring input in context engineering improves signal over noise, addressing a major problem with current AI models [5] - Memory, both short-term and long-term, is crucial for AI, enabling collaboration, remembering conversation history, and effective long-term operations [10][11][12] Knowledge Graphs & Graph RAG - Knowledge graphs provide structured information that complements AI's ability to create and pull from different sources [17] - Graph RAG, which uses graphs as part of the retrieval process, provides more relevant results than vector similarity search by incorporating relationships, nodes, and community groupings [22][23] - Graph RAG enables explainable AI and allows for the implementation of role-based access control, ensuring that only authorized individuals can access specific information [25] Neo4j Solutions & Resources - Neo4j offers a knowledge graph builder, a web application that allows users to upload files and generate knowledge graphs [28] - Neo4j's MCP server is an open-source extension that enables querying knowledge graphs using Cypher, a graph query language [46] - Neo4j provides resources like Graph Academy (free learning resources) and Nodes AI (virtual conference) for learning about graph technology and AI applications [53][54]

Infrastructure Monitoring and Issue Detection - The system proactively monitors application infrastructure, including services, resource metrics (CPU, memory), and HTTP metrics (request error rate, failed requests) [5][8][9] - It analyzes metrics against predefined thresholds to identify affected services, moving beyond simple alert-based systems by analyzing a slice of time to reduce noise from spiky workloads [5][10][11] - The system gathers additional context for suspicious services, including project health, logs, and potentially upstream provider status, to avoid false positives due to high usage or external issues [12][13] Automated Issue Resolution - Upon detecting an issue, the system formulates a detailed plan, leveraging AI to analyze the application architecture, performance data, and errors [14][38] - A coding agent then clones the repository, creates a to-do list based on the plan, implements fixes, and generates a pull request [15] - The coding agent uses Open Code, an open-source AI agent, deployed on a server with necessary tools and Git configured, enabling it to open pull requests [22][23][25][26][27] Durable Workflows and Implementation - The system utilizes durable workflows to manage complex logic and ensure reliability, with automatic retries and caching of successful steps [16][18][19][20] - The workflow involves fetching application architecture, resource metrics, and HTTP metrics via API calls [21][31][32][34] - The system formats the collected information and passes it to the coding agent to generate a fix [33][35][37] Demonstration and Results - A demonstration showcases the workflow, starting from issue detection to the opening of a pull request with proposed changes [6][29][30][40] - The pull request includes a summary of changes, analysis, root causes, and fixes, allowing for review and merging [40][41] - The demonstration highlights a scenario where memory usage is high at 3196% GB out of a maximum of 32 GB, triggering the automated fix [33]