AI Tools & Capabilities Weekly — Week of March 9–15, 2026

This week, the AI landscape jumped forward with major upgrades to business automation, creative tools, and enterprise integration — all pushing agentic AI from experimental phase into practical, everyday use. The biggest theme: agents and AI workflows became both broader (connecting across apps and media types) and deeper (more “do it for me” features and higher security), with large new models ready for real-world scale.

• Google Gemini in Sheets: Now supports full-sheet creation, editing, and advanced data analysis by just describing what you want; nearly matches human experts in spreadsheet performance. Best for: business analysts, finance, and anyone wrangling data. ↗ source
• Nvidia Nemotron 3 Super: A 120-billion-parameter open-weight AI model combining three architectures for high-throughput, low-cost agentic reasoning, beating GPT-OSS and Qwen in many benchmarks. Best for: companies building or running large, complex multi-agent systems. ↗ source
• Gemini Embedding 2: Brings native multimodal (text, image, video, and more) support to enterprise search and retrieval, enabling faster and cheaper vector database tasks. Best for: enterprise data and product teams seeking richer, faster AI-powered search. ↗ source
• Google Maps “Ask Maps”: New Gemini-powered conversational interface so users can plan trips, ask about locations, and get personalized itineraries right inside Google Maps. Best for: anyone planning travel or needing local recommendations. ↗ source
• AI-powered document and slides creation in Google Workspace: Gemini in Docs, Sheets, and Slides now offers AI-powered drafting, content gathering, and editing — just describe your document and get started with one click. Best for: students, marketers, and business professionals. ↗ source
• Anthropic Claude in Excel and PowerPoint: Claude now operates across Microsoft Excel and PowerPoint add-ins, bringing shared context and reusable workflows into popular office apps. Best for: enterprise teams using Microsoft 365. ↗ source
• Wayfair’s AI-powered product catalog: OpenAI models now automate ticket triage and refine millions of product attributes, leading to more accurate ecommerce catalogs. Best for: large retailers and ecommerce teams. ↗ source
• Rakuten’s CI/CD Agent (Codex): Accelerates software delivery by automating code review, deployment pipelines, and cutting incident resolution times in half. Best for: devops teams and software project managers. ↗ source
• NanoClaw-Docker Sandboxes: NanoClaw, with Docker, enables enterprises to run AI agents securely inside isolated sandboxes — limiting risk while allowing real-world actions. Best for: IT admins, security-conscious businesses deploying agentic AI. ↗ source
• OpenAI agent runtime via Responses API: Developers can now run secure, scalable AI agents in containerized environments using OpenAI’s Responses API and shell tool. Best for: developers and businesses building agentic workflows. ↗ source
• Mastercard Agent Pay: First agentic payment system goes live in Singapore: AI agents now autonomously book and pay for rides securely, using Mastercard’s authenticated tokenization model. Best for: fintech innovators, banks, and anyone looking to deploy AI in real-world commerce. ↗ source
• Y Combinator-backed Slate V1: “Swarm-native” coding agent for developers, orchestrating distributed work for large, complex software tasks across multiple AI models. Best for: software engineers and dev shops managing big projects. ↗ source

• Google Gemini Workspace Upgrades: Automated drafting and editing in Docs, Sheets, Slides, streamlining business reports, proposals, and pitch decks. ↗ source
• Wayfair’s AI-enhanced Support & Catalog: Automates ticket triage and improves product data accuracy at ecommerce scale. ↗ source
• Rakuten’s Codex Agent: Automates CI/CD reviews and incident handling, drastically speeding up software release cycles. ↗ source
• OpenAI Responses API Environment: Lets IT leaders deploy scalable, auditable AI agents for business automation and secure workflows. ↗ source
• NanoClaw-Docker Agent Sandboxes: Enables businesses to safely test and deploy autonomous AI agents with strong system isolation. ↗ source
• Manulife’s Agent-Based Workflow Automation: Integrates agentic AI into insurance operations for faster claims and reporting. ↗ source
• SEI & IBM Agentic Finance Automation: Modernizes and automates finance back-office operations with intelligent agents. ↗ source
• Mastercard Agent Pay: Allows AI agents to book and pay for real-world services with authenticated transactions, opening new commerce models. ↗ source

• Gemini in Docs, Slides, and Sheets: Instant drafting, style matching, and content integration from web/email for creative projects and portfolios. ↗ source
• Gemini in Google Sheets: Describe your project or data analysis and let Gemini build the spreadsheet for you, with near-expert layout and completion. ↗ source
• Ask Maps (Gemini in Google Maps): Creators can generate travel guides, curated itineraries, or research locations quickly using conversational prompts. ↗ source
• Anthropic Claude in Excel & PowerPoint: Draft, update, and reuse creative slide decks and data analyses with shared AI context. ↗ source
• Interactive Math & Science Visualizations in ChatGPT: Students, educators, and creators can now explore concepts and formulas in real time through interactive visuals. ↗ source
• Google Gemini Embedding 2: Multimodal embeddings let creators search across text, images, and video for instant inspiration or research in large media archives. ↗ source
• YCombinator-backed Slate V1: Developers and open-source contributors can orchestrate distributed software projects using a swarm of agentic coders. ↗ source
• OpenAI-Promptfoo Acquisition: Security-focused testing tools are improving prompt engineering for creators aiming to build robust generative AI apps. ↗ source

• Mastercard Agent Pay in Singapore: Entrepreneurs can now build autonomous, AI-driven commerce flows through live, tokenized payment APIs. ↗ source
• NanoClaw-Docker Sandboxes: Offers a safe experimentation platform for startups to trial advanced AI agents without risking production data or systems. ↗ source
• Nemotron 3 Super: Open-weight model supports scalable, cost-effective deployment for new AI-driven products and services. ↗ source
• Anthropic Claude for Microsoft Apps: Enables startups to automate client proposals, financial models, and presentations across multiple Microsoft tools. ↗ source
• Google Gemini Embedding 2: Fast-track AI-powered knowledge bases and recommendation systems with multimodal search APIs. ↗ source
• Gemini-powered Google Workspace: Now startups and SMBs can draft legal docs, pitch decks, and data analyses with minimal manual effort. ↗ source
• OpenAI Agent Runtime Platform: Builders can launch secure, containerized AI services leveraging file and stateful operations on demand. ↗ source
• Manulife/SEI AI Automation Models: Case studies demonstrate how founders can design scalable operational flows using intelligent agents as core business components. ↗ source

• 1
  Draft a new report or pitch deck in Google Docs or Slides with Gemini: Click the “Help me write” prompt and enter your outline or main points; Gemini generates a first draft instantly.
• 2
  Ask “Ask Maps” in Google Maps to plan a multi-stop weekend road trip or city tour: Open Maps, tap “Ask Maps,” and describe your desired route or types of stops.
• 3
  Try Gemini in Sheets: Describe a budget or data analysis scenario and let Gemini auto-generate the spreadsheet structure and calculations for you.
• 4
  Use Claude’s Excel add-in: Summarize or analyze a spreadsheet, then reuse findings in a new PowerPoint deck with shared context for seamless presentations.
• 5
  Test Mastercard Agent Pay for an AI-driven real-world purchase (where available): Connect an authenticated agent to a supported service (e.g., airport ride booking) and complete the transaction without human confirmation steps.

• ›Gemini in Google Sheets now rivals human experts for autonomous spreadsheet creation and analysis. ↗ source
• ›Nvidia Nemotron 3 Super open weights set a new standard for high-throughput, cost-efficient AI agent systems. ↗ source
• ›Google’s “Ask Maps” brings Gemini-powered conversational trip planning directly into Maps for US and India users. ↗ source
• ›NanoClaw partners with Docker to sandbox enterprise AI agents for safer, large-scale deployment. ↗ source
• ›Mastercard demonstrates the world’s first live, fully authenticated agentic payment transaction in Singapore. ↗ source
• ›OpenAI’s agent runtime lets developers spin up secure agent containers for real-world workflows. ↗ source
• ›Gemini Embedding 2 unlocks seamless multimodal research and retrieval tasks in the enterprise stack. ↗ source
• ›Slate V1 launches as the first “swarm-native” coding agent to orchestrate big software projects with AI. ↗ source

• Gemini in Google Sheets just achieved state-of-the-art performance.
• Nvidia’s new open weights Nemotron 3 super combines three different architectures to beat gpt-oss and Qwen in throughput
• Google’s Gemini Embedding 2 arrives with native multimodal support to cut costs and speed up your enterprise data stack
• Google Maps Gets Chatty With a New Gemini-Powered Interface
• Gemini burrows deeper into Google Workspace with revamped document creation and editing
• Anthropic gives Claude shared context across Microsoft Excel and PowerPoint, enabling reusable workflows in multiple applications
• Wayfair boosts catalog accuracy and support speed with OpenAI
• Rakuten fixes issues twice as fast with Codex
• NanoClaw and Docker partner to make sandboxes the safest way for enterprises to deploy AI agents
• From model to agent: Equipping the Responses API with a computer environment
• Mastercard brings agentic payments to life in Singapore with DBS and UOB
• Y Combinator-backed Random Labs launches Slate V1, claiming the first ‘swarm-native’ coding agent
• Manulife moves AI agents into core financial workflows
• Agentic AI in finance speeds up operational automation
• OpenAI to acquire Promptfoo
• New ways to learn math and science in ChatGPT