# Mainstage: tech keynote recaps, full data

> Generated from https://mainstage.fru.dev. One block per keynote. Summaries of official keynote videos and recaps; watch the source for exact wording.

## Apple WWDC 2026: Apple WWDC 2026 June 8: Introducing Siri AI and more
https://mainstage.fru.dev/keynotes/apple-wwdc-2026-federighi | 2026-06-08, Cupertino, California | Craig Federighi (Senior Vice President of Software Engineering, Apple) | https://www.youtube.com/watch?v=hF8swzNR1-o
- [33:49] Apple introduced a completely reimagined Siri AI with personal context understanding and onscreen awareness
- [33:49] A dedicated Siri app syncs conversations across devices via iCloud
- [34:35] Visual Intelligence expanded to iPad, Mac and Apple Vision Pro, plus a new Camera app Siri mode on iPhone
- [49:45] Writing Tools can now draft, edit and proofread across apps with tone-aware suggestions
- [30:34] Dictation accuracy improved with automatic capitalization and punctuation
- Siri AI requires iPhone 15 Pro or later, iPad mini (A17 Pro) or later, or an M1 Mac or later
- Siri AI rolls out later in 2026, with availability restricted in the EU and pending regulatory approval in China
- Announced (Preview): Siri AI: Reimagined assistant with personal context and onscreen awareness; beta later in 2026
- Announced (Preview): Dedicated Siri App: Syncs conversations across devices via iCloud
- Announced (Preview): Visual Intelligence expansion: Now on iPad, Mac, Apple Vision Pro; new Camera app Siri mode
- Announced (Preview): Writing Tools integration: Draft, edit, proofread across apps with tone-aware suggestions
- Announced (Preview): Enhanced Dictation: Improved accuracy with automatic capitalization and punctuation
- Source (recap): https://www.apple.com/newsroom/2026/06/apple-introduces-siri-ai-a-profoundly-more-capable-and-personal-assistant/
- Source (video): https://www.youtube.com/watch?v=hF8swzNR1-o

## Microsoft Build 2026: Microsoft Build 2026 | Satya Nadella Opening Keynote
https://mainstage.fru.dev/keynotes/microsoft-build-2026-nadella | 2026-06-02, San Francisco | Satya Nadella (Chairman and CEO, Microsoft) | https://www.youtube.com/watch?v=FFMm454fxNA
- Build 2026 relocated from Seattle to San Francisco for the first time, running June 2-5, 2026.
- [57:41] Microsoft IQ, a new context layer grounding agents in enterprise and world knowledge, is GA.
- [42:58] Work IQ APIs became generally available June 16, giving agents programmatic access to workplace context.
- [104:11] Microsoft Scout, a personal always-on agent built on OpenClaw and Work IQ, launched for Frontier customers.
- [111:50] MAI-Thinking-1, Microsoft AI's first in-house reasoning model, is in private preview on Foundry.
- [85:53] Agent 365 for local agents extends Entra, Defender and Purview to govern agents anywhere they run.
- [129:59] Microsoft Discovery, the agentic science platform, is now generally available.
- [138:23] Majorana 2, Microsoft's quantum chip, hit a 20-second average qubit lifetime with a path to a million qubits.
- Announced (GA): Microsoft IQ: Context layer grounding agents in enterprise and world knowledge
- Announced (Coming): Work IQ APIs: Programmatic access to workplace intelligence, GA June 16
- Announced (Announced): Microsoft Scout: Personal always-on work agent, available to Frontier customers
- Announced (Preview): MAI-Thinking-1: Microsoft AI's first reasoning model, private preview on Foundry
- Announced (Announced): MAI-Image-2.5: Microsoft's first text-to-image and image-to-image models
- Announced (Announced): Agent 365 for local agents: Extends Entra, Defender, Purview to govern local agents
- Announced (GA): Microsoft Discovery: Agentic AI platform for the full science workflow
- Announced (Announced): Majorana 2: Next-gen quantum chip, 1000x more reliable qubit lifetime
- Source (recap): https://blogs.microsoft.com/blog/2026/06/02/microsoft-build-2026/
- Source (video): https://www.youtube.com/watch?v=FFMm454fxNA

## Google I/O 2026: Google I/O '26 Keynote
https://mainstage.fru.dev/keynotes/google-io-2026-pichai | 2026-05-19, Mountain View, California | Sundar Pichai (CEO of Google and Alphabet) | https://www.youtube.com/watch?v=wYSncx9zLIU
- [67:52] Gemini 3.5 Flash launched, combining frontier intelligence with agentic action for coding tasks
- [67:52] Gemini 3.5 Pro is set to roll out the following month
- [17:10] Gemini Omni can generate video from any input and watermarks output with SynthID
- [24:43] Google Antigravity 2.0 launched as an agent-first development platform with a CLI and SDK
- [43:32] AI Mode in Search, upgraded with Gemini 3.5 Flash, passed 1 billion monthly users
- [43:32] The Search box was redesigned to accept text, images, files, video and Chrome tabs
- [35:13] Gemini Spark, a personal AI agent, launched in beta for Ultra subscribers
- [92:47] Android XR audio glasses are coming fall 2026 via Samsung, Warby Parker and Gentle Monster
- Announced (GA): Gemini 3.5 Flash: Frontier intelligence for coding and agentic tasks
- Announced (Coming): Gemini 3.5 Pro: Rolling out next month
- Announced (GA): Gemini Omni: Video generation from any input; SynthID watermarks
- Announced (GA): Google Antigravity 2.0: Agent-first development platform with CLI and SDK
- Announced (GA): AI Mode Search: Upgraded with Gemini 3.5 Flash; 1B+ monthly users
- Announced (Preview): Gemini Spark: Personal AI agent; beta for Ultra subscribers
- Announced (Coming): Google Pics: Image creation and editing tool launching summer 2026
- Announced (Coming): Android XR Audio Glasses: Fall 2026 via Gentle Monster, Warby Parker, Samsung
- Source (recap): https://blog.google/innovation-and-ai/technology/ai/google-io-2026-all-our-announcements/
- Source (video): https://www.youtube.com/watch?v=wYSncx9zLIU

## Google Cloud Next 2026: Google Cloud Next '26 Opening Keynote
https://mainstage.fru.dev/keynotes/google-cloud-next-2026-kurian | 2026-04-22, Las Vegas, Nevada | Thomas Kurian (CEO, Google Cloud), Sundar Pichai (CEO of Google and Alphabet) | https://www.youtube.com/watch?v=11PBno-cJ1g
- [0:12] Thomas Kurian outlined a unified stack of chips, models, agents and security for enterprise AI
- [12:02] The Gemini Enterprise Agent Platform launched GA with Agent Studio, Runtime and Memory Bank
- [39:58] New TPU 8t and TPU 8i chips shipped, with TPU 8t offering 3x the power of Ironwood
- [39:58] NVIDIA Vera Rubin NVL72 platforms were confirmed coming to Google Cloud
- [50:01] Agentic Data Cloud launched with a Knowledge Catalog and cross-cloud lakehouse support
- [64:40] Google completed its acquisition of security company Wiz
- [64:40] New Threat Hunting and Detection Engineering security agents entered preview
- Google Cloud Next '26 drew more than 32,000 attendees to Las Vegas
- Announced (GA): Gemini Enterprise Agent Platform: Agent Studio, ADK, Agent Runtime, Memory Bank, Identity, Registry, Gateway
- Announced (GA): Gemini Enterprise App: Worker interface with Gemini Projects, Deep Think, Microsoft 365 interop
- Announced (GA): TPU 8t: Training-optimized; 3x Ironwood power, 2x performance per watt
- Announced (GA): TPU 8i: Inference-optimized; 80% better performance per dollar
- Announced (GA): Google Cloud Axion N4A: Arm-based processor
- Announced (Coming): NVIDIA Vera Rubin NVL72: Google Cloud among first platforms to offer it
- Announced (GA): Agentic Data Cloud: Knowledge Catalog, Data Agent Kit, Lightning Engine for Spark
- Announced (Preview): Threat Hunting Agent: New Security Operations agent
- Source (recap): https://cloud.google.com/blog/topics/google-cloud-next/next26-day-1-recap
- Source (recap): https://cloud.google.com/blog/topics/google-cloud-next/google-cloud-next-2026-wrap-up
- Source (video): https://www.youtube.com/watch?v=11PBno-cJ1g

## NVIDIA GTC 2026: NVIDIA GTC 2026 Keynote
https://mainstage.fru.dev/keynotes/nvidia-gtc-2026-jensen | 2026-03-16, San Jose | Jensen Huang (Founder and CEO, NVIDIA) | https://www.youtube.com/watch?v=jw_o0xr8MWU
- [0:17] The keynote opened by framing the token as the basic unit of modern AI, at NVIDIA GTC 2026, SAP Center, San Jose.
- [13:47] Huang introduced DLSS 5, using 3D guided neural rendering for real time, photoreal 4K performance on local hardware.
- [51:48] Huang said computing demand has increased by 1 million times in the last two years.
- [70:08] NVIDIA Vera Rubin is a new full stack computing platform: seven chips, five rack scale systems, one supercomputer for agentic AI.
- [98:37] Beyond Vera Rubin, NVIDIA's next major architecture is Feynman, with a new CPU named Rosa after Rosalind Franklin.
- [106:45] OpenClaw, described as the fastest growing open source project in history, lets anyone build a long running AI agent, or claw.
- NVIDIA IGX Thor, an industrial-grade edge computing platform for real time physical AI, is now generally available.
- DGX Spark systems are available now, with clustering of up to four systems announced at the show.
- Announced (Announced): NVIDIA Vera Rubin: Full-stack platform, 7 chips, 5 rack-scale systems, 1 supercomputer for agentic AI
- Announced (Announced): Feynman architecture and Rosa CPU: NVIDIA's next major architecture beyond Vera Rubin
- Announced (Announced): DLSS 5: 3D-guided neural rendering for real-time photoreal 4K on local hardware
- Announced (GA): NVIDIA IGX Thor: Industrial-grade edge platform for real-time physical AI
- Announced (GA): OpenClaw: Open source long-running agent (claw) framework, hands-on at GTC Park
- Announced (Announced): DGX Spark clustering: Clustering of up to 4 DGX Spark systems announced
- Source (recap): https://blogs.nvidia.com/blog/gtc-2026-news/
- Source (video): https://www.youtube.com/watch?v=jw_o0xr8MWU

## CES 2026: NVIDIA Live at CES with CEO Jensen Huang
https://mainstage.fru.dev/keynotes/ces-2026-jensen | 2026-01-05, Las Vegas | Jensen Huang (Founder and CEO, NVIDIA) | https://www.youtube.com/watch?v=0NBILspM4c4
- [86:14] Huang said roughly 10 trillion dollars of the last decade of computing is being modernized to accelerated, AI-based computing.
- [55:31] NVIDIA Rubin, its first extreme-codesigned six-chip AI platform, is now in full production, cutting token cost to about one tenth.
- [21:52] NVIDIA showed a personal AI agent running locally on a DGX Spark desktop supercomputer, embodied in a Reachy Mini robot.
- DGX Spark delivers up to 2.6x performance for large models, with new support for Lightricks LTX-2 and FLUX image models.
- NVIDIA announced Alpamayo, an open portfolio of reasoning vision-language-action models for level-4-capable autonomous driving.
- [40:48] The first car with Alpamayo on NVIDIA DRIVE will be the all-new Mercedes-Benz CLA, with AI-defined driving in the US this year.
- NVIDIA's open model portfolio spans six domains: Clara, Earth-2, Nemotron, Cosmos, GR00T and Alpamayo.
- Announced (GA): NVIDIA Rubin platform: First extreme-codesigned six-chip AI platform, now in full production
- Announced (Announced): NVIDIA Inference Context Memory Storage Platform: AI-native KV-cache tier for long-context inference
- Announced (Announced): Alpamayo (R1, AlpaSim): Open reasoning VLA models and simulation blueprint for autonomous driving
- Announced (Coming): Mercedes-Benz CLA with Alpamayo: First Alpamayo-equipped passenger car, AI-defined driving in the US this year
- Announced (Announced): DGX Spark performance update: Up to 2.6x performance for large models, upcoming NVIDIA AI Enterprise support
- Announced (Announced): NVIDIA Hyperion: Open, modular, level-4-ready platform adopted by automakers and robotaxi providers
- Source (recap): https://blogs.nvidia.com/blog/2026-ces-special-presentation/
- Source (video): https://www.youtube.com/watch?v=0NBILspM4c4

## AWS re:Invent 2025: The Future of Agentic AI Is Here
https://mainstage.fru.dev/keynotes/aws-reinvent-2025-swami | 2025-12-03, Las Vegas | Swami Sivasubramanian (Vice President of Agentic AI, AWS) | https://www.youtube.com/watch?v=prVdCIHlipg
- [1:03] Swami opened by declaring agentic AI as a fundamentally new way of using AI.
- [15:22] Amazon Bedrock AgentCore was built to solve orchestration and reliability challenges for agents.
- [40:59] Amazon SageMaker AI was framed as the answer to the operational overhead of managing agent infrastructure.
- [41:43] Amazon Bedrock now offers Mistral 7B and other open-weight models including Qwen and Kimi.
- [47:30] Amazon Nova Forge, which Garman launched the day before, lets customers build custom frontier models.
- [75:55] Amazon Nova Act reached general availability for building reliable browser automation agents.
- [26:18] Blue Origin's VP of Tech Transformation joined to discuss building agentic systems for space operations.
- Announced (GA): Amazon Nova 2 Sonic: Speech-to-speech model for conversational AI
- Announced (GA): Amazon Nova 2 Lite: Fast, cost-effective reasoning model with extended thinking
- Announced (Preview): Amazon Nova 2 Omni: Multimodal model for text, image, video and speech
- Announced (Announced): Amazon Nova Forge: Program to build custom frontier models on Nova
- Announced (GA): Amazon Nova Act: Service for building UI workflow automation agents
- Announced (Announced): Amazon Bedrock AgentCore: Added quality evaluations and policy controls for agents
- Announced (GA): Amazon Bedrock open-weight models: Adds 18 models including Mistral Large 3 and Ministral 3
- Source (recap): https://aws.amazon.com/blogs/aws/top-announcements-of-aws-reinvent-2025/
- Source (video): https://www.youtube.com/watch?v=prVdCIHlipg

## AWS re:Invent 2025: AWS re:Invent 2025 - Keynote with CEO Matt Garman
https://mainstage.fru.dev/keynotes/aws-reinvent-2025-garman | 2025-12-02, Las Vegas | Matt Garman (CEO, Amazon Web Services) | https://www.youtube.com/watch?v=q3Sb9PemsSo
- [6:22] Garman highlighted how AWS customers like Pfizer chose AWS to transform their business.
- [4:06] AWS Graviton is described as the chip powering the largest share of AWS Cloud compute added.
- [22:25] Trainium3 was announced, designed to make AI training and inference workloads more efficient.
- [91:56] AWS Transform was pitched as AI-powered code and mainframe modernization, cutting execution time up to 80%.
- [122:28] Amazon S3 Vectors reached general availability, scaling to 2 billion vectors per index.
- [119:55] AWS announced Lambda Durable Functions for multi-step workflows lasting seconds to a year.
- [127:18] AWS introduced Database Savings Plans, a new flexible pricing model across AWS databases.
- Announced (GA): AWS Graviton5: Fifth-generation Arm chip, best price-performance for broad workloads
- Announced (Announced): AWS Trainium3: Custom AI chip built for more efficient training and inference
- Announced (GA): AWS Transform: AI-powered code, mainframe and Windows modernization
- Announced (GA): Amazon S3 Vectors: Vector storage scaling to 2 billion vectors per index
- Announced (GA): Lambda Durable Functions: Multi-step serverless workflows lasting up to one year
- Announced (Announced): Database Savings Plans: New flexible pricing model across AWS database services
- Announced (GA): EC2 X8aedz instances: Memory-optimized instances with 5 GHz speeds and 3 TiB memory
- Source (recap): https://aws.amazon.com/blogs/aws/top-announcements-of-aws-reinvent-2025/
- Source (video): https://www.youtube.com/watch?v=q3Sb9PemsSo

## Microsoft Ignite 2025: Microsoft Ignite: Opening Keynote
https://mainstage.fru.dev/keynotes/microsoft-ignite-2025-althoff | 2025-11-18, San Francisco | Judson Althoff (CEO, Microsoft Commercial Business), Asha Sharma (President, Microsoft CoreAI) | https://www.youtube.com/watch?v=TUeET4zY95c
- [3:40] Judson Althoff, CEO of Microsoft Commercial Business, opened the keynote, not Satya Nadella.
- More than 200,000 people registered for Ignite 2025, with 20,000-plus attending in San Francisco.
- [14:51] Work IQ was introduced as a workplace intelligence layer capturing how work actually happens.
- [84:36] Microsoft Agent 365 became generally available for managing and governing agents at scale.
- Agent Mode in Excel lets users choose between Anthropic and OpenAI reasoning models.
- [50:39] Logic Apps connectors expose over 1,400 systems, including SAP and Salesforce, as MCP tools.
- New Frontier-program agents were shown for sales development, workforce insights, and learning.
- Announced (GA): Microsoft Agent 365: Infrastructure to manage and govern AI agents at scale
- Announced (Preview): Sales Development Agent: Autonomous prospecting and outreach agent, Frontier program
- Announced (Preview): Teams Admin Agent: Automates Teams admin tasks like monitoring and provisioning
- Announced (Preview): SharePoint Admin Agent: Monitors inactive sites and manages permissions
- Announced (Preview): Word, Excel, PowerPoint Agents: Agent Mode content generation across Microsoft 365 apps
- Announced (GA): Copilot voice: Natural speech interaction with Copilot for commercial users
- Announced (Preview): New Power Apps maker workspace: Unified AI-powered canvas for app building
- Announced (Preview): Dynamics 365 ERP MCP servers: Connects AI agents to enterprise ERP systems
- Source (recap): https://news.microsoft.com/ignite-2025-book-of-news/
- Source (video): https://www.youtube.com/watch?v=TUeET4zY95c

## KubeCon + CloudNativeCon 2025: Keynote: Welcome + Opening Remarks - Jonathan Bryce & Chris Aniszczyk
https://mainstage.fru.dev/keynotes/kubecon-na-2025-atlanta | 2025-11-11, Atlanta | Jonathan Bryce (Executive Director, Cloud and Infrastructure, Linux Foundation (CNCF)), Chris Aniszczyk (CTO, CNCF) | https://www.youtube.com/watch?v=cQvtT2vRhok
- [0:51] CNCF opened KubeCon NA 2025 in Atlanta by marking 10 years since the foundation's creation
- [3:37] Organizers pointed to landscape.cncf.io to show project maturity levels, with several projects graduating this year
- [7:00] CVS Health was welcomed as a new platinum end-user member of CNCF
- [11:23] CNCF said Kubernetes has powered cloud native infrastructure for a decade and is now the base for AI workloads too
- [31:58] CNCF and OSTIF announced a partnership to fund third-party security audits of core CNCF projects
- [23:56] Kubernetes AI Conformance was framed as a way to certify clusters run AI and ML workloads reliably and portably
- Announced (Announced): Certified Kubernetes AI Conformance Program: v1.0 released, open development at cncf/ai-conformance
- Announced (Announced): CNCF and SlashData cloud native developer survey: Finds 15.6 million cloud native developers worldwide
- Announced (Announced): CNPE Certification: Defines enterprise-scale platform engineering globally
- Source (recap): https://www.cncf.io/announcements/2025/11/11/cncf-launches-certified-kubernetes-ai-conformance-program-to-standardize-ai-workloads-on-kubernetes/
- Source (recap): https://www.cncf.io/announcements/2025/11/11/cncf-and-slashdata-survey-finds-cloud-native-ecosystem-surges-to-15-6m-developers/
- Source (video): https://www.youtube.com/watch?v=cQvtT2vRhok

## Confluent Current 2025: Keynote: Building Intelligent Systems on Real-time Data
https://mainstage.fru.dev/keynotes/confluent-current-2025-kreps | 2025-10-29, New Orleans | Jay Kreps (Co-founder and CEO, Confluent) | https://www.youtube.com/watch?v=qSwVl72etgY
- [1:38] Jay Kreps opened Current 2025 in New Orleans on building intelligent systems with real-time data
- [9:32] Kreps argued that putting MCP in front of systems is not enough on its own, since MCP just exposes access to data
- [25:47] Confluent introduced streaming agents that let LLMs act directly on continuously processed streaming data
- [31:03] Confluent announced a partnership with Anthropic connecting Claude to live enterprise data through MCP
- [83:14] Confluent launched Unified Stream Manager to manage streaming data consistently across the business
- [89:41] A live demo showed the new Real-Time Context Engine serving data to AI systems through a fully managed MCP server
- [96:39] Tableflow was shown converting streaming data into open table formats with a single push-button step
- Announced (GA): Confluent Intelligence: Real-time, trustworthy data platform for production AI
- Announced (GA): Tableflow Delta Lake and Unity Catalog support: Native Databricks integration
- Announced (GA): Confluent Hub 2.0: Redesigned connector marketplace
- Announced (Preview): Real-Time Context Engine: Early access; serves enriched data to AI agents via MCP
- Announced (Preview): Streaming Agents Phase 2: Agent definition, observability, and debugging
- Announced (Announced): Unified Stream Manager: Single control point for streaming data across the business
- Announced (Preview): Queues for Kafka (KIP-932): Dynamic scaling for dedicated clusters
- Announced (Preview): mTLS Authentication: Limited availability for enterprise clusters in December
- Source (recap): https://www.confluent.io/blog/2025-q4-confluent-cloud-launch/
- Source (recap): https://www.confluent.io/blog/introducing-real-time-context-engine-ai/
- Source (video): https://www.youtube.com/watch?v=qSwVl72etgY

## Salesforce Dreamforce 2025: Dreamforce Main Keynote 2025
https://mainstage.fru.dev/keynotes/salesforce-dreamforce-2025-benioff | 2025-10-14, San Francisco | Marc Benioff (Chair and CEO, Salesforce) | https://www.youtube.com/watch?v=Gce2DhADtK4
- [2:11] Benioff opened Dreamforce 2025 by declaring Salesforce is entering the age of the Agentic Enterprise
- [18:52] Salesforce introduced Agentforce Voice, giving agents predictable, low-latency voice capability
- [22:18] Benioff said Slack has become an agentic operating system for connecting people, agents, and data
- [24:05] Salesforce launched Agentforce Vibes, a low-code, conversational way to build AI agents
- [27:36] A new version of Tableau, driven by Tableau Semantics, was shown translating business language into data
- [56:06] Data 360 was introduced to give agents intelligent context across unstructured and structured data
- [31:48] Customer stories featured Dell, FedEx, PepsiCo, Pandora, and Williams-Sonoma using Agentforce
- Announced (GA): Agentforce 360: Conversational builder, hybrid reasoning, native voice, observability
- Announced (GA): Agentforce Voice: Low-latency native voice for agents
- Announced (Announced): Agentforce Vibes: Low-code AI development experience
- Announced (GA): Data 360: Intelligent Context plus Tableau Semantics for unstructured data
- Announced (GA): Slack-First Apps and Channel Expert Agent: Agentic search and apps built into Slack
- Announced (GA): Model Context Protocol support: Third-party AI tool integration via MCP
- Source (recap): https://www.salesforce.com/news/press-releases/2025/10/13/agentic-enterprise-announcement/
- Source (recap): https://www.salesforce.com/news/stories/dreamforce-2025-live-blog/
- Source (video): https://www.youtube.com/watch?v=Gce2DhADtK4

## OpenAI DevDay 2025: OpenAI DevDay 2025: Opening Keynote with Sam Altman
https://mainstage.fru.dev/keynotes/openai-devday-2025-altman | 2025-10-06, San Francisco | Sam Altman (CEO, OpenAI) | https://www.youtube.com/watch?v=hS1YqcewH0c
- DevDay 2025 was held October 6, 2025 at Fort Mason in San Francisco.
- [3:46] The Apps SDK launched in preview, letting developers build apps that run inside ChatGPT.
- [16:52] AgentKit was introduced as a complete toolkit: Agent Builder, Connector Registry and ChatKit.
- [30:41] Codex was cited alongside image generation and Sora as part of OpenAI's product lineup.
- [31:22] Codex is now generally available with a Slack integration and a new Codex SDK.
- [33:16] The Codex Slack integration lets teams tag @Codex directly in a channel or thread.
- Apps SDK pilot partners included Booking.com, Canva, Coursera, Figma, Expedia, Spotify and Zillow.
- Announced (Preview): Apps SDK: Open standard, built on MCP, for building apps inside ChatGPT
- Announced (Announced): Apps in ChatGPT: First-party apps from Booking.com, Canva, Coursera, Figma, Expedia, Spotify, Zillow
- Announced (Announced): AgentKit: Agent Builder, Connector Registry and ChatKit for building agents
- Announced (GA): Codex: General availability with Slack integration and Codex SDK
- Announced (Announced): Codex SDK: Embed the Codex CLI agent into custom workflows, TypeScript first
- Source (recap): https://openai.com/index/introducing-agentkit
- Source (recap): https://openai.com/index/introducing-apps-in-chatgpt
- Source (recap): https://openai.com/index/codex-now-generally-available
- Source (video): https://www.youtube.com/watch?v=hS1YqcewH0c

## MongoDB.local 2025: MongoDB.local NYC 2025 Keynote
https://mainstage.fru.dev/keynotes/mongodb-local-nyc-2025-ittycheria | 2025-09-18, New York City | Dev Ittycheria (President and CEO, MongoDB (at time of keynote)) | https://www.youtube.com/watch?v=HzKLWv66N9k
- [1:20] Dev Ittycheria opened MongoDB.local NYC 2025 on what the ideal database for the AI era looks like
- [4:29] MongoDB said it now serves nearly 60,000 organizations, with over 70% of the Fortune 500 as customers
- [20:13] MongoDB 8.2 launched, described as the company's most feature-rich and performant release yet
- [46:11] MongoDB announced Search and Vector Search for self-managed Community and Enterprise Server deployments
- [72:56] MongoDB highlighted its Voyage AI embedding models, noting Anthropic recommends Voyage models to customers
- [17:19] MongoDB introduced the Application Modernization Platform (AMP), an AI-driven way to modernize legacy systems
- [29:00] McKesson shared how it uses MongoDB to secure its pharmaceutical supply chain under DSCSA requirements
- Announced (GA): MongoDB 8.2: Most feature-rich and performant MongoDB release yet
- Announced (Preview): Search and Vector Search for self-managed deployments: Public preview for Community Edition and Enterprise Server
- Announced (Preview): Voyage AI embedding models and rerankers: For building trustworthy, reliable AI applications
- Announced (Preview): MongoDB Application Modernization Platform (AMP): AI-driven legacy system modernization
- Announced (Announced): New release cadence and 5-year Long-Term Support: Longer support window for MongoDB releases
- Source (recap): https://www.mongodb.com/company/blog/events/local-nyc-2025-defining-ideal-database-for-ai-era
- Source (video): https://www.youtube.com/watch?v=HzKLWv66N9k

## Meta Connect 2025: Meta Connect 2025 Keynote
https://mainstage.fru.dev/keynotes/meta-connect-2025-zuckerberg | 2025-09-17 | Mark Zuckerberg (Founder and CEO, Meta) | https://www.youtube.com/watch?v=Q9eLemEd9FY
- Zuckerberg said Ray-Ban Meta glasses have become the number one selling AI glasses in the world, with millions sold.
- [6:31] Ray-Ban Meta (Gen 2) launched with nearly double the battery life of Gen 1, up to eight hours, starting at 379 dollars.
- [13:21] Oakley Meta Vanguard, a new Performance AI glasses line for high-intensity sports, was announced with EssilorLuxottica, at 499 dollars.
- [19:20] Meta Ray-Ban Display, AI glasses with a full-color in-lens display, was unveiled starting at 799 dollars including the Neural Band.
- [19:47] Meta Neural Band is an EMG wristband, built on research with nearly 200,000 participants, that reads muscle signals to control glasses.
- Meta Ray-Ban Display began shipping September 30 at select US retailers including Best Buy, LensCrafters and Sunglass Hut.
- Announced (Coming): Meta Ray-Ban Display: 799 dollars with Neural Band, shipped September 30 at select US retailers
- Announced (Coming): Meta Neural Band: EMG wristband bundled with Meta Ray-Ban Display
- Announced (Coming): Oakley Meta Vanguard: 499 dollars, pre-order at announcement, launched October 21
- Announced (GA): Ray-Ban Meta (Gen 2): 379 dollars, available at announcement in most countries
- Source (recap): https://about.fb.com/news/2025/09/meta-ray-ban-display-ai-glasses-emg-wristband/
- Source (recap): https://about.fb.com/news/2025/09/oakley-meta-vanguard-performance-ai-glasses-sports/
- Source (recap): https://about.fb.com/news/2025/09/ray-ban-meta-gen-2-better-battery-life-video-capture/
- Source (video): https://www.youtube.com/watch?v=Q9eLemEd9FY

## Databricks Data + AI Summit 2025: Ali Ghodsi, Co-founder and CEO, Databricks kicks off Data + AI Summit 2025
https://mainstage.fru.dev/keynotes/databricks-data-ai-summit-2025-ghodsi | 2025-06-11, San Francisco | Ali Ghodsi (Co-founder and CEO, Databricks) | https://www.youtube.com/watch?v=PFVJL7_W4dI
- [4:26] Databricks now counts more than 15,000 customers running on the platform across every major cloud
- [6:22] Ghodsi said hundreds of customers already run mission-critical production workloads on Databricks
- [15:06] Unity Catalog was framed as the default governance layer customers pick for their data and AI assets
- [17:36] AI/BI Genie, which lets people query data in plain English, is now used by 81% of customers
- [18:34] Ghodsi said 98% of customers are already using Databricks' AI assistant capabilities
- Databricks unveiled Databricks One, a no-code interface for business users to use dashboards and Genie
- Databricks and Google Cloud partnered to bring Gemini models natively into the Databricks platform
- Announced (Preview): Databricks One: Private preview, code-free interface for business users
- Announced (GA): Lakebase: Managed Postgres-compatible transactional database for AI agents
- Announced (GA): Lakeflow: End-to-end data engineering and pipeline ingestion
- Announced (GA): AI/BI Genie: Natural-language queries; Deep Research mode coming soon
- Announced (GA): Databricks Apps: Build and deploy governed data intelligence apps
- Announced (Preview): Agent Bricks: Production AI agents tuned on customer data
- Announced (Preview): Lakeflow Designer: No-code ETL pipeline builder
- Announced (Preview): Databricks Free Edition: Free access to data and AI skills for learners
- Announced (Announced): Spark Declarative Pipelines: Open-source declarative batch and streaming ETL
- Announced (Announced): Lakebridge: Automated migration from legacy data warehouses
- Announced (Announced): Gemini on Databricks: Strategic AI partnership with Google Cloud
- Source (recap): https://www.databricks.com/events/dataaisummit-2025-announcements
- Source (recap): https://www.databricks.com/company/newsroom/press-releases/databricks-unveils-databricks-one-new-experience-bring-data-and-ai
- Source (recap): https://www.databricks.com/company/newsroom/press-releases/databricks-announces-strategic-ai-partnership-google-cloud-bring
- Source (video): https://www.youtube.com/watch?v=PFVJL7_W4dI

## Apple WWDC 2025: WWDC 2025 Keynote
https://mainstage.fru.dev/keynotes/apple-wwdc-2025-federighi | 2025-06-09, Cupertino, California | Craig Federighi (Senior Vice President of Software Engineering, Apple) | https://www.youtube.com/watch?v=0_DjDdfqtUE
- [10:58] iOS 26 introduces a new Liquid Glass design language across Apple's operating systems
- [4:09] Apple Intelligence gets deeper system integration a year after its debut
- [4:29] Visual Intelligence lets users search what's on their screen, with ChatGPT integration for follow-up questions
- [28:08] Live Translation arrives in Messages, FaceTime and Phone with on-device processing
- [20:13] CarPlay Ultra gets a Liquid Glass redesign with customizable layouts and widgets
- [35:52] A new unified Games app gathers Apple Arcade and every game in one hub
- [30:48] Apple Music adds AutoMix for automatic DJ-style transitions between songs
- Announced (Preview): iOS 26: Liquid Glass redesign, expanded Apple Intelligence; dev beta now, GA fall 2025
- Announced (Coming): Games app: Unified hub for Apple Arcade and all games, GA fall 2025
- Announced (Coming): Live Translation: On-device translation in Messages, FaceTime, Phone; GA fall 2025
- Announced (Coming): Visual Intelligence: Screen-based search with ChatGPT integration; GA fall 2025
- Announced (Coming): CarPlay Ultra: Liquid Glass redesign, new widgets and Live Activities; GA fall 2025
- Announced (Coming): Apple Music AutoMix: Automatic DJ-style song transitions; GA fall 2025
- Source (recap): https://www.apple.com/newsroom/2025/06/apple-elevates-the-iphone-experience-with-ios-26/
- Source (video): https://www.youtube.com/watch?v=0_DjDdfqtUE

## Snowflake Summit 2025: Snowflake Summit 2025 Opening Keynote With Sridhar Ramaswamy, Sam Altman, And Sarah Guo
https://mainstage.fru.dev/keynotes/snowflake-summit-2025-ramaswamy | 2025-06-02, San Francisco | Sridhar Ramaswamy (CEO, Snowflake), Sam Altman (CEO, OpenAI), Sarah Guo (Founder and Managing Partner, Conviction) | https://www.youtube.com/watch?v=gJf39VG87O8
- [15:25] Snowflake highlighted deeper Apache Iceberg support, used by companies like Booking.com, via Catalog Linked Databases
- [28:59] Sam Altman and Conviction founder Sarah Guo joined Ramaswamy on stage for a fireside chat on AI and agents
- [30:24] OpenAI and Snowflake said their partnership has driven down the cost of accessing frontier models
- [26:05] Discussion turned to natural-language interfaces for data, foreshadowing the Snowflake Intelligence launch
- [36:37] Altman called an AI agent completing complex scientist-level tasks unassisted a moment that felt like AGI
- Snowflake was named the official data collaboration partner of the LA 2028 Olympic Games
- NYSE President Lynn Martin joined Ramaswamy for a separate keynote conversation
- Announced (Preview): Snowflake Intelligence: Natural-language chat with enterprise data via ai.snowflake.com
- Announced (Preview): Cortex AISQL: Multimodal AI processing through SQL syntax
- Announced (GA): Snowflake Openflow: Managed multimodal data ingestion built on Apache NiFi, GA on AWS
- Announced (Preview): Data Science Agent: Automates steps of the ML workflow
- Announced (Coming): dbt Projects on Snowflake: Native dbt development inside Snowsight UI
- Announced (Preview): Snowflake Semantic Views: Reusable business semantics for analytics
- Announced (GA): Standard Warehouse Generation 2: 2.1x faster core analytics performance
- Announced (Preview): Snowflake Adaptive Compute: Private preview of self-tuning compute
- Announced (Coming): Cortex Knowledge Extensions: GA planned soon for Cortex AI observability
- Source (recap): https://www.snowflake.com/en/blog/announcements-snowflake-summit-2025/
- Source (video): https://www.youtube.com/watch?v=gJf39VG87O8

## Anthropic events 2025: Code with Claude Opening Keynote
https://mainstage.fru.dev/keynotes/anthropic-code-with-claude-2025 | 2025-05-22, San Francisco | Mike Krieger (Chief Product Officer, Anthropic), Dario Amodei (Co-founder and CEO, Anthropic) | https://www.youtube.com/watch?v=EvtPBaaykdo
- [27:19] Mike Krieger opened Code with Claude, Anthropic's first developer conference, in front of a live audience
- [29:35] Dario Amodei announced Claude Opus 4 and Claude Sonnet 4 were releasing live across all Anthropic products
- [34:28] Claude 4 models add near-instant responses alongside an extended thinking mode for harder problems
- [37:18] Amazon's Alexa team used Claude to prototype new assistant functionality ahead of the Alexa+ launch
- [40:59] A new code execution tool in the API lets Claude write and run code directly during a conversation
- [44:29] Claude Code can now run inside GitHub workflows using the Claude Code SDK
- [52:55] Anthropic said it created MCP as a universal translator between AI models and external tools, and that Microsoft is building MCP into its Build keynote plans
- Announced (GA): Claude Opus 4: Flagship model, leads SWE-bench at 72.5%
- Announced (GA): Claude Sonnet 4: Mid-tier model, state of the art 72.7% on SWE-bench
- Announced (GA): Claude Code: Expanded to general availability with IDE integrations
- Announced (GA): Code execution tool: API tool letting Claude execute code directly
- Announced (Preview): Extended thinking with tool use: Beta reasoning mode with web search during processing
- Announced (Announced): Model Context Protocol adoption: Anthropic formalizing MCP as an open standard
- Source (recap): https://www.anthropic.com/news/claude-4
- Source (video): https://www.youtube.com/watch?v=EvtPBaaykdo

## Google I/O 2025: Google I/O '25 Keynote
https://mainstage.fru.dev/keynotes/google-io-2025-pichai | 2025-05-20, Mountain View, California | Sundar Pichai (CEO of Google and Alphabet) | https://www.youtube.com/watch?v=o8NiE3XMPrM
- [2:17] Google now processes over 480 trillion tokens per month across its products
- [2:17] The Gemini app passed 400 million monthly active users
- [46:04] AI Mode in Search rolled out to all US users with deeper reasoning via Gemini 2.5
- [84:02] Veo 3 generates video with native, synchronized audio
- [84:02] Imagen 4 became the latest image generation model available in the Gemini app
- [71:06] Gemini 2.5 Pro gained an experimental Deep Think reasoning mode
- [71:06] Agent Mode in the Gemini app can handle multi-step tasks like apartment hunting
- Google Beam, an AI 3D video communication platform, is coming later in 2025 with HP as first hardware partner
- Announced (Coming): Google Beam: AI-powered 3D video communications platform; first devices via HP
- Announced (GA): Gemini Live: Camera and screen-sharing on Android, rolling out to iOS
- Announced (Coming): Agent Mode (Gemini app): Experimental; automates tasks like apartment hunting
- Announced (GA): AI Mode (Search): Available now in the US with Gemini 2.5 integration
- Announced (Preview): Gemini 2.5 Pro Deep Think: Experimental reasoning mode
- Announced (GA): Veo 3: Video generation with native audio, in the Gemini app
- Announced (GA): Imagen 4: Latest image generation model
- Announced (GA): Flow: AI filmmaking tool for cinematic content
- Source (recap): https://blog.google/technology/ai/io-2025-keynote/
- Source (video): https://www.youtube.com/watch?v=o8NiE3XMPrM

## Microsoft Build 2025: Microsoft Build 2025 | Satya Nadella Opening Keynote
https://mainstage.fru.dev/keynotes/microsoft-build-2025-nadella | 2025-05-19, Seattle | Satya Nadella (Chairman and CEO, Microsoft) | https://www.youtube.com/watch?v=ceV3RsG946s
- [2:05] Satya Nadella framed the event around building an open agentic web at scale.
- [3:20] GitHub Copilot now has 15 million developers using it.
- [7:23] GitHub Copilot coding agent is a full agentic partner built into GitHub.
- [21:17] Microsoft 365 Copilot Tuning lets customers train agents on their own company data.
- [55:04] Microsoft announced broad first-party support for Model Context Protocol (MCP).
- [62:31] Windows AI Foundry unifies the AI developer lifecycle for training and inference on Windows.
- [82:34] NLWeb was introduced as an open project giving websites a conversational, MCP-native interface.
- [106:51] Microsoft Discovery was unveiled as an agentic AI platform to accelerate scientific R&D.
- Announced (Announced): GitHub Copilot coding agent: Asynchronous agentic coding partner built into GitHub
- Announced (Announced): Windows AI Foundry: Unified platform for the AI developer lifecycle on Windows
- Announced (GA): Azure AI Foundry Agent Service: Orchestrates multi-agent workflows; Semantic Kernel + AutoGen in one SDK
- Announced (Preview): Microsoft Entra Agent ID: Assigns unique identities to agents to prevent agent sprawl
- Announced (Announced): Model Context Protocol support: First-party MCP support across GitHub, Foundry, Copilot Studio, Windows 11
- Announced (Announced): NLWeb: Open project giving websites a conversational, MCP-native interface
- Announced (Announced): Microsoft Discovery: Agentic AI platform for accelerating scientific R&D
- Announced (Announced): Microsoft 365 Copilot Tuning: Low-code agent creation using a company's own data
- Source (recap): https://blogs.microsoft.com/blog/2025/05/19/microsoft-build-2025-the-age-of-ai-agents-and-building-the-open-agentic-web/
- Source (video): https://www.youtube.com/watch?v=ceV3RsG946s

## Computex 2025: NVIDIA CEO Jensen Huang Keynote at COMPUTEX 2025
https://mainstage.fru.dev/keynotes/computex-2025-jensen | 2025-05-19, Taipei | Jensen Huang (Founder and CEO, NVIDIA) | https://www.youtube.com/watch?v=TLzna9__DnI
- [14:21] Huang told a crowd of more than 4,000 at Taipei Music Center that AI data centers are, in fact, AI factories.
- NVIDIA detailed huge new AI installations from CoreWeave, Oracle, Microsoft and xAI, calling them gigantic factory investments.
- [26:21] NVIDIA and Foxconn Hon Hai are deepening their partnership to build an AI factory supercomputer for Taiwan, including TSMC.
- Huang announced NVLink Fusion, letting hyperscalers build semi-custom compute with NVIDIA's NVLink interconnect.
- [25:01] DGX Spark, a personal AI supercomputer for developers, is now in full production and ships in a few weeks.
- NVIDIA RTX PRO Servers, for agentic AI, are now in volume production as part of the Enterprise AI Factory design.
- [79:10] NVIDIA partnered with Google DeepMind and Disney to build Newton, a physics training engine for robotics.
- [98:38] NVIDIA revealed Constellation, a new NVIDIA office in Taiwan for its growing Taiwan workforce.
- Announced (GA): NVIDIA Blackwell: Engine powering the AI ecosystem, one architecture from cloud to edge
- Announced (Announced): NVLink Fusion: New architecture letting hyperscalers build semi-custom compute with NVLink
- Announced (Coming): DGX Spark: Personal AI supercomputer, full production, available in a few weeks
- Announced (Announced): DGX Station: Up to 20 petaflops from a wall socket, runs 1 trillion parameter models
- Announced (GA): NVIDIA RTX PRO Servers: Enterprise servers for agentic AI, now in volume production
- Announced (Announced): NVIDIA AI Data Platform: Reference design with RTX 6000 PRO Blackwell Server Edition GPUs
- Announced (Announced): Newton physics engine: Physics training engine for robotics, with Google DeepMind and Disney
- Announced (Announced): Isaac GR00T-Dreams / GR00T N1.5: Blueprint and foundation model to speed humanoid robot development
- Announced (Announced): NVIDIA Constellation: New NVIDIA office in Taiwan for its growing Taiwan workforce
- Source (recap): https://blogs.nvidia.com/blog/computex-2025-jensen-huang/
- Source (video): https://www.youtube.com/watch?v=TLzna9__DnI

## Google Cloud Next 2025: Google Cloud Next 25 Opening Keynote: The new way to cloud
https://mainstage.fru.dev/keynotes/google-cloud-next-2025-kurian | 2025-04-09, Las Vegas, Nevada | Thomas Kurian (CEO, Google Cloud) | https://www.youtube.com/watch?v=Md4Fs-Zc3tg
- [2:53] Google Cloud shipped over 3,000 product advancements in the prior year
- [33:22] Vertex AI usage grew 20x year over year
- [16:19] 7th-generation Ironwood TPUs deliver 42.5 exaflops per pod, a 10x jump from the prior generation
- [16:19] A4 and A4X VMs made Google Cloud the first provider offering both NVIDIA B200 and GB200
- [46:40] The open-source Agent Development Kit (ADK) lets developers build agents in under 100 lines of code
- [46:40] The Agent2Agent (A2A) protocol launched with over 50 partners for multi-agent interoperability
- [18:44] Gemini 2.5 Pro entered public preview as a new thinking model for precision tasks
- [58:18] Google Agentspace expanded with Chrome Enterprise integration and a new Agent Gallery
- Announced (GA): Ironwood TPUs (7th gen): 42.5 exaflops per pod, 10x improvement
- Announced (GA): A4 & A4X VMs: NVIDIA B200 and GB200, first cloud to offer both
- Announced (Coming): NVIDIA Vera Rubin GPUs: Up to 15 exaflops FP4 inference per rack
- Announced (GA): Cloud WAN: Up to 40% performance improvement, 40% TCO reduction
- Announced (Preview): Gemini 2.5 Pro: Public preview thinking model for precision tasks
- Announced (Announced): Agent Development Kit (ADK): Open-source; build agents in under 100 lines of code
- Announced (Announced): Agent2Agent (A2A) Protocol: Open protocol for multi-agent ecosystems, 50+ partners
- Announced (Announced): Google Agentspace: Expanding with Chrome Enterprise integration, Agent Gallery
- Source (recap): https://cloud.google.com/blog/topics/google-cloud-next/welcome-to-google-cloud-next25
- Source (video): https://www.youtube.com/watch?v=Md4Fs-Zc3tg

## NVIDIA GTC 2025: NVIDIA Founder and CEO Jensen Huang's Keynote at GTC 2025
https://mainstage.fru.dev/keynotes/nvidia-gtc-2025-jensen | 2025-03-18, San Jose | Jensen Huang (Founder and CEO, NVIDIA) | https://www.youtube.com/watch?v=_waPvOwL9Z8
- NVIDIA Blackwell is in full production, delivering 40x the performance of Hopper for AI training and inference.
- [85:28] Blackwell Ultra, the next evolution of the Blackwell AI factory platform, arrives in the second half of 2025.
- [104:04] NVIDIA Vera Rubin architecture systems are set to arrive in the second half of 2026 for major performance gains.
- [68:15] NVIDIA Dynamo, open source software, is described as the operating system of an AI factory for scaling reasoning models.
- DGX Spark and DGX Station, personal AI supercomputers built on the Grace Blackwell platform, were announced.
- The Llama Nemotron family of open reasoning AI models was announced for building agentic AI platforms.
- Isaac GR00T N1 was announced as an open, customizable foundation model for humanoid robot reasoning.
- [43:04] Cosmos World Foundation Models were released, an open customizable reasoning model for physical AI.
- Announced (GA): NVIDIA Blackwell: In full production, 40x Hopper performance for training and inference
- Announced (Coming): Blackwell Ultra: Next Blackwell platform evolution, arriving second half of 2025
- Announced (Coming): Vera Rubin: Next architecture, systems arriving second half of 2026
- Announced (Announced): NVIDIA Dynamo: Open source AI factory operating system for scaling reasoning models
- Announced (Announced): DGX Spark and DGX Station: Personal AI supercomputers on Grace Blackwell platform
- Announced (Announced): Llama Nemotron: Open reasoning AI model family for agentic AI
- Announced (Announced): Isaac GR00T N1: Open foundation model for generalized humanoid reasoning
- Announced (Announced): Cosmos World Foundation Models: Open customizable reasoning models for physical AI
- Announced (Announced): Newton: Open source physics engine for robotics, with Google DeepMind and Disney Research
- Announced (Announced): Spectrum-X and Quantum-X: Silicon photonics networking switches, fewer lasers, more power efficient
- Source (recap): https://blogs.nvidia.com/blog/nvidia-keynote-at-gtc-2025-ai-news-live-updates/
- Source (video): https://www.youtube.com/watch?v=_waPvOwL9Z8