Deep-dive into AgentKit—Agent Builder, ChatKit, and integrated evals & tracing—with a semi‑truck maintenance demo, exportable code, and practical tips to build, deploy, and optimize agentic workflows at scale.

AgentKit Guide: Design, Deploy & Optimize Agentic Workflows at Scale

Orchestrating Agents at Scale: Building, Deploying, and Optimizing Agentic Workflows with AgentKit

Click, connect, create. Learn how to quickly design and deploy enterprise-grade agents with a new suite of agentic platform tools.

Orchestrating Agents at Scale

Explore how Cursor evolved from next‑action autocomplete to autonomous coding agents — leveraging Tab (400M requests/day), low‑latency UX, and online RL with ~30‑minute updates to boost developer productivity.

Cursor's Context Engineering: From Autocomplete to Coding Agents

Context Engineering & Coding Agents with Cursor

Cursor has grown from next action prediction to fully autonomous coding agents. Learn how they’ve built their agent harness and tools, approached context engineering, and enabled support for running many agents in parallel.

Learn how finance teams use Perplexity’s Spaces, Labs, and Comet to pull SEC tables, automate research, and build repeatable workflows—saving analyst-hours and improving decisions.

Find Alpha Faster: Perplexity's Spaces, Labs & Comet for Finance Teams

Finding Alpha with Perplexity: How Finance Teams Use Perplexity Spaces, Labs, and Comet

Join Perplexity's Enterprise for Finance lead to learn how Finance teams use Perplexity Spaces, Labs, and Comet to find Alpha.

We'll share how you can use Perplexity to:
• Market Thesis Validation: Critically evaluate a market trend to build, support, or challenge investment theses. Deliver data-driven insights and scenario analyses that enable financial institutions to make informed capital allocation and portfolio management decisions
• Evaluate Public Companies: Evaluate public companies with consensus estimates, watchlists, and AI-enabled equities screener
• Identify M&A targets: Identify relevant M&A targets for a given company and provide strategic rationale, synergies analysis, and other considerations
• and more

Finding Alpha with Perplexity: How Finance teams use Perplexity Spaces, Labs, and Comet

Discover how Omi uses photo‑real 3D digital twins and on‑demand NVIDIA GPU rendering to deliver consistent, studio‑quality product visuals at scale—fast and cost‑efficient.

Scale Product Visuals with 3D Digital Twins and NVIDIA GPUs

Omi: Scaling Brand Visuals with 3D Digital Twins

Omi enables brands to create high-quality, on-brand product visuals in minutes—no traditional photo shoots required.

Hear from Omi Co-Founder and CEO, Hugo Borensztein, on how Omi leverages NVIDIA technology and partnerships through the #NVIDIAInception program to unlock new possibilities for rapid, creative, and scalable product content through its virtual product studio.

Learn more about the NVIDIA Inception program for startups: https://nvda.ws/46W6aJ1
Learn more about Omi: https://omi.so/

A front-line report from The Soldier’s Diary CZ examining why nearly 80% of German anti-aircraft (Flak) crews died in WWII—technical analysis and eyewitness detail.

Inside the 80% Death Rate: Why German Flak Crews Rarely Survived WWII

Why 80% of German Anti-Aircraft Gunners Did Not Survive World War II

Německé protiletadlové dělostřelectvo, známé jako Flak, bylo jedním z nejviditelnějších symbolů obrany Třetí říše. Jeho kanóny – od lehkých 20mm až po obávané 88mm – byly rozmístěny ve městech, továrnách a na frontových liniích s cílem zastavit neutuchající spojenecké bombardování. To, co pro civilní obyvatelstvo představovalo znak odporu, bylo pro dělostřelce rozsudkem smrti: bojovali vždy odhalení, bez ochrany, stali se snadnými cíli pro bombardéry, stíhačky a tanky. Tato zranitelnost vysvětluje, proč téměř 80 % posádek válku nepřežilo.

Kanóny Flak nebyly zaměřeny pouze na oblohu – používaly se také jako protitankové zbraně na východní frontě a v Africe. Jejich palebná síla byla ničivá, ale každý výstřel prozradil jejich pozici a přivolal okamžitou odvetu. Ve městech jako Berlín, Hamburk či Essen byly baterie nepřetržitě bombardovány, zatímco u Kurska či Stalingradu byly obsluhy uvězněny v přímých bojích proti vlnám sovětských obrněnců. Dělostřelci prožívali nekonečné dny hladu, chladu, únavy a neustálého strachu, vědomi si toho, že každý útok může být jejich poslední.

S postupem války byli veteráni nahrazováni adolescenty z Volkssturmu a mladými pomocníky Flakhelfer, kterým bylo sotva 15 nebo 16 let. Postrádali výcvik a byli posíláni do městských front nebo do betonových věží. Jejich životy mizely mezi těžkými granáty, výbuchy a nekonečnými nocemi pod nepřátelskou palbou. Výsledek byl ničivý: desítky tisíc dělostřelců zahynuly tiše, bez slávy a uznání. Historie Flaku je příběhem kolektivní oběti – nejtragičtější tvář německé obrany ve druhé světové válce.

Proč 80 % protiletadlových dělostřelců nepřežilo druhou světovou válku?

A practical playbook for adding persistent memory to AI agents using ADK and Vertex AI Memory Bank — generation, retrieval, TTL tuning, and cost savings.

Build Memory-Powered AI Agents with ADK & Vertex

How to build AI agents with memory — a complete playbook

Join Google's Sita Lakshmi Sangameswaran and Kimberly Milam, the tech lead for Vertex AI Memory Bank, as they walk you through the complete playbook for building agents that remember. You will learn the difference between volatile and persistent memory, how to generate, store, and retrieve information across conversations, and even how to customize your agent's memory with a time-to-live (TTL).

Resources:
Code from this tutorial → https://goo.gle/3ISrbw7  
ADK Memory docs → https://goo.gle/3IRLY2O 
Learn more about Vertex AI Memory Bank → https://goo.gle/472zTkJ 

Chapters:
0:00 - The problem with forgetful AI agents
0:34 - What you'll learn in this video
1:15 - Agentic memory concepts
2:58 - Introduction to ADK memory services
3:51 - Code walkthrough: Build a simple conversational agent
5:20 - Source code for different memory services in ADK
10:20 - Define Vertex AI Memory Bank
12:19 - How Memory generation works in InMemoryMemory service
17:40 - How to use Vertex AI Memory Bank with ADK
22:45 - Agent Engine SDK to generate memories
24:20 - Memory consolidation
26:58 - Automate memory generation with callbacks
28:44 - Search and use memory in your agent
34:10 - Customize Memory storage and extraction
38:51 - Set TTL for memories
41:58 - Eng Talk: Community questions
44:08 - Recap

Watch more Getting started with Agent Developer Kit (ADK) → https://goo.gle/42m5Em8 
Subscribe to Google for Developers → https://goo.gle/developers 

#ADK  #VertexAI #GenerativeAI 

Speaker: Sita Lakshmi Sangameswaran, Kimberly Milam 
Products Mentioned: Google AI

How to build AI agents with memory

An inside look at NVIDIA and OpenAI’s historic AI infrastructure plan: 10 gigawatts (~4–5M GPUs), NVIDIA’s up-to-$100B staged investment, the engineering challenges, partners, and implications.

Inside NVIDIA–OpenAI’s 10GW AI Buildout and $100B Investment

NVIDIA and OpenAI Announce ‘the Biggest AI Infrastructure Project in History’

In a live broadcast with CNBC’s Jon Fortt, NVIDIA CEO Jensen Huang, OpenAI CEO Sam Altman, and OpenAI President Greg Brockman described a new strategic partnership to fuel OpenAI’s growth — and enable AI at scale for virtually every industry and user.

Through the partnership, OpenAI will deploy at least 10 gigawatts of NVIDIA systems for OpenAI’s next-generation AI infrastructure, including the NVIDIA Vera Rubin platform. NVIDIA also intends to invest up to $100 billion in OpenAI progressively as each gigawatt is deployed.

Learn more: https://blogs.nvidia.com/blog/openai-nvidia/

Discover Claude Sonnet 4.5 — our best coding model yet — plus Claude Code 2.0 (VS Code extension, terminal UI), checkpoints, context editing, and file-based memory.

Claude Sonnet 4.5 & Claude Code 2.0 — Major Coding Model, Dev Tools, and API Upgrades

Claude Coded: Sonnet 4.5, Claude Code 2.0, and more

We're covering everything new with Claude for developers, including the launch of Claude Sonnet 4.5, major updates to Claude Code, powerful new API capabilities, and exciting features in the Claude app.

Join the Claude Developer Discord - https://anthropic.com/discord
Check out the Claude Code Docs - https://docs.claude.com/en/docs/claude-code/overview 
Check out the Memory Cookbook - https://github.com/anthropics/claude-cookbooks/blob/main/tool_use/memory_cookbook.ipynb 

00:06 Claude Sonnet 4.5 - Leading on SWE-bench and can focus for 30+ hours on complex tasks
01:03 Claude Code VS Code Extension - Now available in beta on the VS Code Marketplace
01:30 Claude Code 2.0 - Refreshed terminal UI
01:45 Claude Code Checkpoints - Roll back to previous states with /rewind
02:23 Usage Tracking - Monitor your usage in real-time with /usage command
02:35 Claude API - Context editing and memory tools for building sophisticated agents
03:37 Claude Agent SDK - Build your own agents with the tools that power Claude Code
03:59 Claude File Creation - Generate Excel, PowerPoint, Word, and PDF files in the Claude app

Claude Coded: Sonnet 4.5, Claude Code 2.0, and more.

See how Apps in ChatGPT connects ChatGPT to third‑party apps for real‑time data, actions, and personalized workflows — rolling out to Free, Go, Plus, and Pro users outside the EU.

ChatGPT Integrations: Chat with Apps for Real‑Time Actions

Apps in ChatGPT — Chat with apps in ChatGPT

You can now chat with apps in ChatGPT.

Apps in ChatGPT are starting to roll out today to Free, Go, Plus, and Pro users outside of the EU. More pilot partner apps and availability coming soon.

Apps in ChatGPT

Discover Agent Builder: a drag-and-drop visual tool for assembling multi-step AI workflows. Learn node types, a travel assistant demo, and how to export to ChatKit or Agents SDK.

Visual Agent Builder — Build Agentic Workflows Without Code

Intro to Agent Builder

Christina Huang from OpenAI guides you through Agent Builder—a new visual tool to create agentic workflows. 

Drag and drop nodes, connect tools, and publish your workflows with ChatKit and the Agents SDK. 

https://platform.openai.com/agent-builder

See how Google’s AI Mode, powered by Gemini 2.5, uses text, voice, images and uploads to answer complex questions—and lets you follow up for clarity.

Google AI Mode (Gemini 2.5): Turn Confusion into Clear Answers

Lost the plot? Just ask Google to get the tea. ☕️

Trying to wrap your head around a new topic? Looking for specific recommendations? Need help deciding between two products? Just ask. Type, talk, snap a photo, or upload an image. AI Mode uses advanced multimodal capabilities to deeply understand your question, no matter how you ask.

AI Mode is our most powerful AI search. Ask whatever's on your mind to get an AI powered response, and explore further with follow-up questions and helpful web links. AI Mode uses Gemini 2.5’s advanced reasoning, thinking and multimodal capabilities to help with even your toughest questions.

www.google.com

#AIMode #JustAskGoogle 

Subscribe to our Channel: https://www.youtube.com/google
Find us on X: https://twitter.com/google
Watch us on TikTok: https://tiktok.com/@google
Follow us on Instagram: https://www.instagram.com/google
Join us on Facebook: https://www.facebook.com/Google

A firsthand investigative profile of Gabriel Peterson and the niche world of duckling racing—the rise to fame, the accident known as "the quack," and the comeback behind the film.

AI and Technology News

Latest