In partnership with
AIHOY and happy weekend, AInauts!
Maybe you missed some of last week's most interesting AI news, tools, and practical tricks, or perhaps you have only just joined us. No worries.
We have rounded up our highlights from the past week:
To wrap things up, we have also collected the most important Quick News for you. That way, every relevant development is only a click away. Ready? Let's go!
AI made PMs faster. Multiplayer mode is still broken.
A PM can summarize research, draft a PRD, and mock up a prototype before lunch. The hard part starts when the team has to decide what actually gets built.
Jira Product Discovery gives product teams one place to capture insights, prioritize ideas with consistent frameworks, and build living roadmaps stakeholders can rally around.
And because it’s connected to Jira, the context behind every decision stays with the work—so developers and their agents know not just what to build, but why.
AI helps PMs move faster. Jira Product Discovery helps the whole team build with confidence.
GPT-6 Astra is by far the best model you can get right now. But it did not simply become smarter. It operates browsers and apps much more reliably, works for longer stretches, and increasingly turns software into a workspace for agents. That means the bottleneck is more often sitting in front of the screen 😉. Give Astra a task your previous model failed at and see what happens.
200+ Proven Ways to Make Money With AI in 2026
The next wave of millionaires will be people who figured out how to make AI work for them.
The window to get ahead is still open. But not for long.
Here are 200+ proven ways to make money with AI in 2026.
Sign up for Superhuman AI, the free daily newsletter read by 1M+ professionals, and get instant access to all 200+ ways to profit from AI this year.
Astra's computer skills are most impressive in the desktop app. We give you three prompts: the first breaks your workweek into routine, simplify, and keep. The second turns a vague wish into a clearly scoped assignment with permissions and completion criteria. The third checks your setup, usage limits, and reset times.
While OpenAI celebrates the next leap in productivity, the safety questions are getting louder. Agent swarms found ways out, models gamed evaluations, and the pressure to grow toward an IPO is increasing.
The practical rule remains dry but important: separate accounts for agents, only the permissions they actually need, and approval gates before sending, buying, or publishing. A digital employee needs boundaries, not blind trust.
Meta is launching Muse in the US as a personal agent inside a messenger-style interface. It plans trips, writes emails, shops, and works in its own secure cloud environment. Payments and sensitive steps come back to you for approval.
We find the strategy compelling: if Meta scales this cleanly, the AI employee for everyone could suddenly feel as normal as a voice message.
We taught Claude a new trick: with our skill, it can ask GPT, Gemini, or another model directly from the current chat.
Claude remains the orchestrator, while the other models provide second opinions. That is especially useful for important writing, decisions, and code reviews.
OpenAI released ChatGPT Images 2.5 along with the API models Flare and Sunburst. We put versions 2 and 2.5 through the same visual tests. The new generation produces more realistic textures, better lighting, and preserves more of the original during edits.
Sunburst was the most precise in our test, while Flare is optimized for speed. The difference is visible, but strong references and clear change requests still matter more than ten adjectives.
AI News Quickie: The HAI-lights from the Industry
AI never sleeps!
New features to try, important subscription changes, and a little glimpse of the future. Let's start with the major providers.
🎯 OpenAI and ChatGPT: More Help for Everyday Work
ChatGPT Images 2.5 generates faster and edits more precisely. On mobile, Sketch lets you scribble an idea and turn it into a finished image.
ChatGPT Work can now pick up your writing style, including typical phrasing, sign-offs, and other habits from connected apps such as Gmail, Drive, Slack, or SharePoint.
ChatGPT Voice is getting Astra for more demanding questions. Plus users receive up to three hours of GPT-Live-1 per day; afterward, there is no automatic switch to the Mini version. Availability and limits depend on your plan.
The ChatGPT Library now also includes Box, Dropbox, and SharePoint. You can open and analyze connected files next to the chat without uploading them again each time.
You can now share files and folders in ChatGPT and grant viewing or editing access. Useful for teams, but note that files in someone else's shared folder still belong to that folder's owner.
Deep Research in ChatGPT Work researches across the web, files, and connected apps and produces editable results with sources.
The Data Agent in ChatGPT Work connects to company data and turns questions into analyses and dashboards.
ChatGPT for Financial Services connects Astra with financial data and Excel, Word, and PowerPoint templates. For now, the offering is limited to eligible financial institutions.
OpenAI reports an internal rate of 3.1 agent workdays per human workday. That measures deployed compute work, not automatically three times the productivity.
At the same time, OpenAI's research chief warns in An Alien Mind about progress whose consequences society and safety systems may not be prepared for.
🧠 Anthropic and Claude: Subscription Frustration and Safety Questions
Anthropic models three economic scenarios through 2030, ranging from an internet-like boost to rapid growth with major pressure on jobs and wages. These are scenarios, not a weather forecast for your workplace.
Anthropic examines four cyber testing incidents in which Claude accessed real external systems. Three were already known; the combined analysis and fourth case are new. A misconfigured test environment had left internet access open.
A new Anthropic test shows how accurately Claude can infer locations from sparse clues. Safeguards are intended to prevent this from becoming a tool for surveillance or targeting.
A dispute over Claude Max subscriptions has expanded into a class action, but there is no ruling yet. Anthropic says the 5x and 20x figures apply per session; additional weekly limits still remain.
💎 Google and Gemini: Desktop, Daily Life, and Music
The Gemini app for Windows opens with Alt+Space and adds a workspace for multi-step Spark tasks.
Google's Dreambeans app creates personalized daily stories in the US from Calendar, Gmail, Photos, Search, YouTube, and Gemini.
With Lyria 3.5 in Gemini, you can create music with or without vocals, choose a genre and duration, or start from a template. It is available worldwide for uses such as background music or a catchy birthday greeting.
The AlphaGenome Atlas maps nine billion DNA variants. Google's methane model simultaneously finds significantly more emissions sources from space.
🧰 Microsoft, Meta, and AI Helpers for Daily Life and Business
Microsoft's MAI-Image-2.6 combines multiple reference images and makes targeted edits. Both variants are available in the MAI Playground.
Meta Muse is launching in the US through its app and WhatsApp. Many everyday tasks are free, with subscriptions available for heavier use.
xAI had Grok Bot audit procurement for unused software subscriptions and costly supplier contracts and reports more than $100,000 in direct savings.
DeepSeek V4.1 Flash understands text and images and lowers the cost of using it in your own tools and automations.
Slack is using Slackforce Surfaces to turn chats and data into interactive reports, dashboards, and microsites. An enabled Slackbot is required.
Apple's Siri AI launches with limits: English only at first, daily usage caps, and no initial access on iPhone, iPad, and Watch in the EU.
The new Health app with Apple Intelligence is designed to make sleep, fitness, and long-term health data easier to understand. It will arrive later this year and launches in US English.
💼 Creative Tools, Media, and What Comes Next
Suno v6 lets you change individual song sections or lyrics with an instruction instead of rebuilding the entire song. The Mini version is available to everyone.
Universal Music and ElevenLabs are also planning a licensed music platform.
The free DaVinci Resolve 21.1 connects AI assistants to the video editor. It also adds expanded photo editing and more than 100 new tools and controls.
In Adobe Premiere 26.5, you can generate video and sound effects directly on the timeline. Paper Edit builds a rough cut from selected transcript sentences.
Prime Video now adapts lip movements to human-dubbed voices.
Made it. See you in the next issue.
Reto & Fabian from AInauten
