Hey Rocky, great to have you here. >> Hey, so can you tell me what's new with voice. >> Absolutely.Voice is now built right into our chat, so you get a live transcript as we talk. Plus, I can show you things like maps, weather, and more in real time. >> Okay. Okay.Can you show me a map of the best bakeries in the Mission District. [music] >> Sure. Here's a map of the top bakeries in the Mission District.As you'd expect, Tartine is right up there as a favorite. >> Tartine. Okay.What kind of pastries do they ...

Agent Memory Patterns & Context Engineering - Context engineering is both an art and a science, requiring judgment to decide what matters most while employing concrete patterns and methods for systematic context management [1] - Modern LLMs perform based on the context provided, making context engineering a broader discipline than prompt engineering or retrieval, encompassing prompt engineering, structured output, RAG, state and history management, and memory [1] - Core strategies for effective context management include reshape and fit (context trimming, compaction, summarization), isolate and route (offloading context to sub-agents), and extract and retrieve (memory extraction, state management, memory retrieval) [1] - Short-term memory (in-session techniques) focuses on maximizing the context window during active interaction, while long-term memory (cross-session) builds continuity across sessions [1] - Context management challenges include context burst (sudden token spikes), context conflict (contradictory information), context poisoning (incorrect information propagation), and context noise (redundant tool definitions) [2] Techniques and Solutions - Solutions involve managing context efficiently using techniques like trimming, compaction, state management, and memories, moving beyond prompt engineering [3] - AI agents can be grouped into RAG-heavy assistants, tool-heavy workflows, and conversational concierges, each with different context profiles [3] - Prompting best practices include being explicit and structured, giving room for planning and self-reflection, and avoiding conflicts in tool definitions [3] - Engineering techniques include context trimming (dropping older turns), context compaction (dropping tool calls from older turns), and context summarization (compressing prior messages into structured summaries) [3][4] - Memory shapes can range from simple JSON notes to complex paragraphs, with extraction using memory tools and state management involving defining state objects [26][27][28][29] Best Practices and Evaluation - Best practices in agent memory design include understanding the typical context, deciding when and how to remember and forget, and continuously cleaning and consolidating memories [31][32] - Evaluation involves running evals with and without memory, building memory-specific evals for long-running tasks, and finding the right heuristics for context engineering techniques [36][37][38][39][40] - Strategies for keeping memory fresh include temporal tags and weighted decay to manage stale memories and prioritize recent information [46][47][48][49] - Scaling agent memory systems involves considering whether to use a retrieval-based approach (scaling vector databases) or a summarization approach (scaling data storage) [51][52][53][54][55]

Algorithms are hard to visualize and for me it's literally impossible. I have affantasia which means that I can't form mental images at all. But today I'm going to walk you through how I use codeex openi's agentic coding tool to build my very own algorithms visualizer website so I can finally see what my brain is unable to generate by itself.Step one, tell what to build. I asked Codeex to come up with a list of common algorithms and create a website with their visualizations. That's it.No need to specify la ...

When we announced that we would have this pilot going, then we had more than twice the requests to join, then we actually had seats. We've been partners with OpenAI for uh about a year. Um we're mainly working currently with the SHA GPT rolling that out across the enterprise.uh but we're also exploring the use of APIs. >> We put quite a lot of uh licenses available in the organization and we stimulate experimentation as much as possible. when we did the introduction then everyone was uh only allowed to join ...

AI Adoption & Literacy - Philips is elevating AI literacy across its 70,000 employees to ensure everyone knows how to use AI [2] - Philips initiated a journey with OpenAI a few months ago, deploying it among a few thousand users to observe adoption and benefits in daily practices [2] - Philips' XCO members were personally trained to use the enterprise chat GPT tool [3] - A summer challenge was launched to encourage employees to propose AI applications for the company, with licenses awarded to the best ideas [3][4] Healthcare Efficiency & Innovation - Philips aims to reduce the administrative burden on healthcare practitioners, freeing up time for patient care [5] - The goal is to fully utilize AI's potential in healthcare, going "for the moon" and "Mars" [5] - Clinicians spend an equivalent amount of time on administration as they do on saving lives, highlighting the need for AI to improve efficiency [4] AI Integration - AI is already embedded in many Philips products, utilizing traditional AI and machine learning [2] - Philips views AI as a tool for playing, working, and innovating [1]

AI Integration & Product Development - Notion 通过与 OpenAI 的合作，将 AI 技术融入产品，提升了产品的功能和用户体验 [1][2] - Notion AI 使工作效率提高 50% [3] - Notion AI 连接到用户的 Notion 工作区以及 GitHub、Confluence 和 Slack 等其他平台 [3] - Notion 利用 OpenAI 的嵌入技术，实现了快速和高质量的信息检索 [4] User Engagement & Satisfaction - 使用 Notion AI 的客户活跃度比不使用的客户高出约 66.67% (twothirds) [4] - 75% (3/4) 的用户表示 AI 已经成为他们使用 Notion 不可或缺的一部分，并且不会放弃使用 [5] Competitive Advantage & Confidence - OpenAI 认为 Notion 的问答产品非常出色，增强了 Notion 对其产品世界一流的信心 [5]

We created this atmosphere of uh being in a safe place uh to learn and to use AI. We've been users of Open AI Chat GPT for the last couple of years. Uh we have deployed the the product to actually initially 3,000 employees across many different countries and then very quickly we jumped to 11,000.We're very proud of what we have there. We have a 83% uh weekly recurrence in our in our people saving around 3 hours average per employee and we are measuring efficiencies of around 80% in many cases. >> We have de ...

Long recognized as one of the world’s most innovative financial institutions, BBVA is now applying AI at global scale: as a new way of working and a core capability across the organization. We sat down with Antonio Bravo, Global Head of Data & AI and Elena Alfaro, Head of Global AI Adoption, to hear how the bank is scaling AI adoption across teams and business units. Read more: https://openai.com/index/bbva-2025/ ...

Product Update - Introduction of GPT-5.1 and new personalization controls for all users [1] - Eight unique chat styles are available in ChatGPT, including three new options [1] - New chat styles are accessible across all ChatGPT models [1] Chat Style Personalization - Users can now personalize ChatGPT with eight distinct chat styles: Default, Professional, Candid, Quirky, Friendly (formerly Listener), Efficient (formerly Robot), Nerdy, and Cynical [1]

Code review of two comment mark. >> Hey everyone, I'm Roma >> and I'm Maya. >> Codex needs to do two things really well to be an effective coding teammate.First, it needs to work with all your tools and second, it also needs to plug into all of your team workflows. Code review is one of the most important workflows for any engineering team and we want to help there as well. With GPT5 and now GP5 codeex, we train these models specifically to find bugs and investigate some issues.Maya, why don't you tell us m ...