In partnership with
Hello from The AI Night,
Today in AI:
Claude is Now Inside The Excel
Google Introduced Personal Intelligence in AI Mode Search
Subagents are now available in cursor 2.4
Image source: Claude blog
Here's the deal: Anthropic has expanded Claude in Excel from a limited beta to all Pro, Max, Team and Enterprise subscribers. The add in embeds Claude directly into Microsoft Excel's sidebar, where it can read, analyze, modify and create workbooks.
The Breakdown:
Powered by Opus 4.5, reads multi tab workbooks with nested formulas and cross sheet dependencies
Provides cell level citations for every explanation users can verify logic directly
Updates assumptions while preserving formula structures, highlights all changes with explanations
Debugs #REF!, #VALUE!, and circular reference errors by tracing to source
Now supports pivot tables, charts, and file uploads (added November 2025)
Keyboard shortcut: Ctrl+Option+C (Mac) / Ctrl+Alt+C (Windows)
Works within existing enterprise security frameworks, chat history not persisted between sessions
The bigger picture: Anthropic is directly challenging Microsoft Co-pilot on its home turf. For finance teams and analysts who live in spreadsheets, this creates a viable alternative with stronger citations and transparency turning Excel from a productivity tool into an AI native workspace.
Image Source: Google blog
Here's the deal: Google launched "Personal Intelligence" for AI Mode in Search, allowing the assistant to pull context from your Gmail and Google Photos to deliver personalized recommendations. The feature is rolling out to Google AI Pro and Ultra subscribers in the U.S. as an opt in Labs experiment.
The Breakdown:
AI Mode can now reference email confirmations (flights, hotel bookings) and photo history to tailor search responses
Powered by Gemini 3, Google states it doesn't train directly on your inbox or photo library
Currently limited to personal accounts ,Workspace business, enterprise and education users excluded
Users control connections via Search personalization settings and can disable anytime
The bigger picture: This moves Google Search from answering generic queries to acting as a context aware personal assistant. For users deep in the Google ecosystem, it reduces friction in planning, shopping and discovery. It also raises the competitive bar personalization tied to first party data across Gmail, Photos and Search is a competitive edge only Google can leverage at scale. Expect other AI assistants to pursue similar cross app integrations.
Image Source: Cursor Tweet
Here's the deal: Cursor released version 2.4, introducing subagents, independent agents that run in parallel to handle discrete subtasks with their own context, prompts and tool access. The update also adds a skills system and native image generation.
The Breakdown:
Subagents execute specialized work streams simultaneously, reducing context bloat in main conversations
Default subagents handle codebase research, terminal commands and parallel workflows automatically
Skills are defined in SKILL.md files dynamic, procedural instructions agents discover when relevant
Image generation uses Google Nano Banana Pro, outputs save to assets/ folder
Enterprise only "Cursor Blame" tracks AI vs. human authorship per line, linking code to the conversation that produced it
Agents can now ask clarifying questions mid task while continuing work in the background
The bigger picture: This release pushes AI coding tools toward genuine orchestration not just chat based assistance. Subagents and skills let developers modularize agent behavior, while Cursor Blame addresses a growing enterprise concern, auditability of AI generated code. Competitors will face pressure to match these workflow primitives.
Turn AI into Your Income Engine
Ready to transform artificial intelligence from a buzzword into your personal revenue generator?
HubSpot’s groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.
Inside you'll discover:
A curated collection of 200+ profitable opportunities spanning content creation, e-commerce, gaming, and emerging digital markets—each vetted for real-world potential
Step-by-step implementation guides designed for beginners, making AI accessible regardless of your technical background
Cutting-edge strategies aligned with current market trends, ensuring your ventures stay ahead of the curve
Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.
What else you need to know:
Sakana AI secured a strategic partnership and investment from Google following its Series B round to build products using Gemini and Gemma models and deploy AI in regulated Japanese industries.
Google DeepMind launched an agricultural landscape layer in Google Earth for portions of Asia-Pacific that uses ML and satellite imagery to map individual field boundaries and calculate acreage.
Anthropic released Petri 2.0 with 70 new behavioral audit scenarios and realism mitigations that reduced eval-awareness by 47% in Claude models while adding benchmark results for recent frontier models.
Alibaba's Qwen open sourced Qwen3-TTS, a text-to-speech model series (1.7B and 0.6B parameters) supporting voice cloning, voice design, and natural language style control across 10 languages.
Kilo deployed a stealth reasoning model codenamed “Giga Potato” from a Chinese open-source lab that outperforms open-weight models on long context coding tasks with a 256k context window and 32k output limit.
That’s it for today’s edition of The AI Night.
Our goal is to cut through the noise, surface what actually changed, and explain why it matters.
If this was useful, you’ll get the same signal here tomorrow.