In this newsletter:

• Gemini exports responses as Word, Excel, and PDF files
• Mistral launches cloud coding agents for async tasks
• ChatGPT adds physical security keys for account protection

Plus, you’ll find new AI tools and this week’s top AI news headlines!

📄 Gemini Now Exports Responses as Downloadable Word, Excel, and PDF Files

Google rolled out an update letting Gemini generate downloadable files directly in chat, eliminating the need to copy-paste responses and manually format them.

What's new:

• Available now for all Gemini app users globally on web and mobile
• Type a query and ask the chatbot to respond to the preferred file type
• Supported formats: .pdf, .docx, .xlsx, .csv, LaTeX, plain text (TXT), rich text format (RTF), markdown (MD)
• If linked to a Workspace account, can export directly to Google Drive as a Doc, Sheet, or Slide
• Gemini handles formatting automatically for Word documents, Excel sheets, and PDFs
• Can add photos of handwritten notes to Gemini and convert them to downloadable files
• Can generate diagrams and other visual data with proper formatting

How it works:

• Google: "Instead of copying, pasting, and reformatting, this update will allow you to easily move your work into different applications."
• Visuals in outputs look presentable across all file formats
• Example use: Ask Gemini to generate a road trip itinerary in PDF and .docx formats with decent formatting

What's coming:

• Beta version of Google app for Android shows upcoming Proactive Assistance feature
• Will suggest ideas of what Gemini can do without asking, such as setting up a quiz to help revise for an exam

Why it matters:

The file export feature positions Google as the productivity-first AI by making Gemini outputs immediately usable in workplace and school workflows without manual reformatting, pressuring OpenAI to either match the feature or watch users switch to Gemini for work tasks.

Google is betting that reducing friction between AI generation and real-world deliverables matters more than raw model intelligence, a strategic shift that could redefine how people choose AI tools based on output format rather than conversation quality.

👀 Read more about Gemini file export updates!

🔧 Mistral Launches Cloud Coding Agents That Run Tasks While You're Away

Mistral AI released remote coding agents in Mistral Vibe that run in the cloud, work in parallel, and notify you when done, powered by the new Mistral Medium 3.5 model.

What's new:

• Available now on Pro, Team, and Enterprise plans
• Start cloud agents from Mistral Vibe CLI or Le Chat
• Sessions run in the cloud while you step away, and can run multiple in parallel
• Local CLI sessions can be teleported to the cloud with session history, task state, and approvals carried across
• Each session runs in an isolated sandbox
• The agent can open a pull request on GitHub and notify you when work is done
• Plugs into GitHub for code and PRs, Linear and Jira for issues, Sentry for incidents, Slack or Teams for reporting
• Fits high-volume work like module refactors, test generation, dependency upgrades, CI investigations, bug fixes

Mistral Medium 3.5:

• Dense 128B model with 256k context window
• Handles instruction-following, reasoning, and coding in a single set of weights
• Released as open weights under a modified MIT license
• Self-hosting possible on as few as four GPUs
• Reasoning effort configurable per request for quick chat reply or complex agentic run
• Vision encoder trained from scratch to handle variable image sizes and aspect ratios
• SWE-Bench Verified: 77.6%, ahead of Devstral 2 and Qwen3.5 397B A17B
• τ³-Telecom: 91.4
• API pricing: $1.5 per million input tokens, $7.5 per million output tokens
• Replaces Devstral 2 in Vibe CLI, becomes the default model in Le Chat
• Available on NVIDIA build.nvidia.com and as NVIDIA NIM containerized inference microservice

Work mode in Le Chat:

• New agentic mode for complex tasks powered by Mistral Medium 3.5
• Cross-tool workflows: catch up across email, messages, and calendar in a single run; prepare for a meeting with attendee context, news, and talking points
• Research and synthesis: dive into the topic across the web, internal docs, and connected tools, produce a structured brief or report
• Triage inbox and draft replies, create Jira issues from discussions, send summary to team on Slack
• Sessions persist longer than a typical chat reply
• Connectors on by default rather than chosen manually
• Every action is visible with tool calls and thinking rationale
• Asks for explicit approval before sensitive tasks like sending a message, writing a document, or modifying data

Why it matters:

Mistral's cloud-hosted agents shift the competitive battleground from which AI writes better code to which AI requires the least human supervision, forcing Anthropic and OpenAI to either match the hands-free workflow or lose developers who can't justify babysitting sessions.

The open-weights release under a modified MIT license reveals Mistral's strategy to win through enterprise self-hosting rather than cloud subscriptions, undercutting competitors who lock models behind APIs while betting companies will pay for deployment flexibility over convenience.

👀 Read more about Mistral remote coding agent!

🔐 ChatGPT Gets Physical Security Keys to Stop Account Hijacking

OpenAI launched Advanced Account Security with partner Yubico, offering hardware security keys that protect ChatGPT accounts from phishing attacks through physical authentication.

What's new:

• Advanced Account Security (AAS) is available to all ChatGPT users who want extra protection
• Two physical YubiKeys available: YubiKey C NFC and YubiKey C Nano
• Small hardware device plugs into your computer's USB port
• You need the physical key in your hand to log into your ChatGPT account
• Even if someone steals your password, they can't access your account without the physical key
• Designed for people handling sensitive work: journalists, researchers, political activists, government officials, business users
• Warning: If you lose the key, OpenAI cannot recover your account or conversations - everything is lost permanently

Why this exists:

• Hackers are increasingly targeting ChatGPT users to steal sensitive conversations
• Anthropic announced the cybersecurity model Mythos several weeks ago
• OpenAI announced a new digital defense framework

Why it matters:

OpenAI is treating ChatGPT conversations as valuable enough to need the same security level as bank accounts, admitting that people are storing sensitive information that criminals want to access.

The move creates a new category of "high-security AI users" who need physical keys to protect their chat history, directly competing with encrypted messaging apps like Signal for journalists and political dissidents who previously wouldn't trust cloud-based AI with their most sensitive work.

👀 Read more about OpenAI account security measures!

My Latest LinkedIn & X/Twitter Posts:

• 10 powerful ways to use ChatGPT Deep Research (view post)
• 9 ChatGPT prompts to extract strategies and find insights (view post)
• How to utilize NotebookLM to boost your productivity (view post)
• 10 ChatGPT prompt templates to get better results faster (view post)
• 9 ChatGPT prompt frameworks for expert-level results (view post)

In partnership with Kit:

​Monetize Your Audience With My Favorite All-in-One Tool​

As a creator, your time should be spent doing what you love, not juggling a dozen tools just to run your business.

Kit gives you everything you need in one place:

✅ Build and grow your email list (I use Kit for my newsletter)
✅ Easily monetize with paid newsletters and digital products
✅ Automate your emails with triggers and custom workflows
✅ Track what’s working and optimize with powerful insights

It’s not just email. It’s your entire creator business, simplified.

Join a thriving community of successful creators.

​Use Kit for Free — and Start Building Smarter​

New AI Tools to Boost Your Productivity:

• ​Claude Security: Helps teams find and fix security issues faster.
• ​Bolta: Automates social media planning, writing, and publishing.
• ​Free Fact Checker: Fact-checks AI responses and online content instantly.
• ​LovedByAI: Helps businesses get discovered in AI search results.
• ​ApparenceKit: Template for building mobile apps.
• ​Lens: Identifies objects from uploaded images.
• ​AI Beat Maker: Generates music beats from prompts.
• ​SecondGlance: Analyzes and improves written replies for clarity.
• ​AI Music: Generates songs and music videos from prompts.
• ​Nya AI: All-in-one workspace for chat, notes, and collaboration.
• ​IG Follower Export: Exports Instagram follower data for leads.
• ​Kael: Finds jobs and generates tailored applications.
• ​QuikWit: Builds chatbots for real estate sales and support.
• ​MiaDance: Generates images from text prompts.
• ​HeySprite: Writes content automatically.

This Week's Top AI News Headlines:

• Google’s Gemini AI Assistant is Rolling Out to Millions of Vehicles, Bringing Voice-Controlled AI Features into Cars for Navigation, Messaging, and In-Car Tasks (View Article)
• Spotify Introduces “Verified by Spotify” Badge to Help Artists and Creators Prove Authenticity and Build Trust With Listeners (View Article)
• YouTube Tests AI-Powered Search Feature that Delivers Guided Answers and Curated Video Results Instead of Traditional Listings (View Article)
• X Launches Rebuilt Ad Platform Powered by AI to Improve Targeting, Campaign Performance, and Advertiser Tools (View Article)
• Writer, an Enterprise AI Platform Startup, Launches Autonomous Agents that Act Without Prompts to Compete With Amazon, Microsoft, and Salesforce (View Article)
• Stripe Expands Link Digital Wallet to Support AI Agents that Can Automatically Shop and Complete Purchases on Behalf of Users (View Article)
• Apple Sees Unexpected Surge in Mac Demand Driven by AI Workloads as Users Upgrade Hardware for Local AI Processing (View Article)
• OpenAI Restricts Access to “Cyber” Model After Criticizing Anthropic’s Limits on Mythos, Highlighting Growing Tensions Over AI Safety Controls (View Article)
• Meta Says its Business AI Now Handles 10 Million Weekly Conversations as Companies Rapidly Adopt AI for Customer Interactions (View Article)
• Elon Musk Testifies that xAI Trained Grok Using OpenAI Models, Raising Questions About Data Sources and AI Development Practices (View Article)
• Salesforce Is Crowdsourcing its AI Roadmap by Letting Customers Shape Product Development and Influence Future AI Features (View Article)
• SoftBank Plans Robotics Company to Build AI Data Centers and Targets Massive $100 Billion IPO as Demand for Compute Infrastructure Surges (View Article)
• Suno Bets $2.5 Billion on AI-Generated Music as the Startup Pushes to Make Synthetic Songs a Mainstream Industry (View Article)

Work With Me:

• Paid Partnerships: I’ll promote your product on my LinkedIn, X/Twitter or newsletter. Fill out form to get more info.
• Consulting: I’ll help you grow on LinkedIn & X/Twitter or use AI to boost marketing and sales. ​Book consulting call​.

If you enjoy this newsletter, please forward it to your friends and colleagues.

Follow me on LinkedIn and X/Twitter to see my latest posts.

Have a wonderful week!

Andrew Bolis