👨‍🚀 Brilliant! Google hacks your workspace

Hi AInauts,

Welcome to the latest issue of your favorite newsletter!

This week, we discovered three updates that are too good not to share right away. At least one of them will make your everyday life easier, we promise.

Included: A conversation decoder hack for perfect prompts, Google's research autopilot for Gmail & Drive, plus a Chinese model that redefines open source.

• 🎯 Prompt hack: How to turn your best AI chats into one-shot prompts

• 🥇 Google unlocks Workspace Turbo: Finally more time for deep work

• 🔥 Better than GPT and Claude—and even open source!

Let's go!

🎯 Prompt Hack: How to turn your best AI chats into one-shot prompts

Does this sound familiar? You've spent half an hour tinkering with ChatGPT or Claude until you finally get the result you want. And then you think: "That's exactly how I want it next time—only without all the back and forth!"

Read on, we've found a solution: Advanced chat reverse engineering!

The idea is simple but effective: instead of laboriously crafting the perfect prompt by hand, you let the AI become the prompt architect itself.

AI chatbots are experts at recognizing patterns. When you ask them to figure out the best way to get to a destination, you're tapping into that exact strength.

At the end of a successful conversation, simply say:

Analyze our conversation and create the optimal prompt that would have produced the final result in a single pass—including all contextual information and structural specifications!

The result is an optimized meta prompt that you can easily save and retrieve when needed. You'll get the best results if you ask the AI to include "confidence markers." This allows you to see immediately where fine-tuning is still needed.

To make sure this sticks, don't wait until your next chat to try it out. Instead, find a conversation in your chats now and apply it to that 😁.

🥇 Google unlocks Workspace Turbo: Finally more time for deep work

Who would have thought that the annoying game of "searching through all emails" would one day be a thing of the past? AI does it with a single prompt!

Gemini Deep Research can now search your entire Google Workspace history, link it to current web data, and deliver you a ready-to-use report. In a matter of minutes.

Here's how you can apply it directly:

1. Open Gemini → Tools → Deep Research

2. Enable Workspace sources (Gmail, Drive, Chat)

3. Enter promptly, done

Google is gradually rolling out the feature for all Gemini users. In a corporate environment, the administrator must enable access in the Workspace settings ("Allow access to Deep Research"). If you don't see it yet, be patient or pester your administrator...

... or use ChatGPT, Claude, or Microsoft Copilot. These chatbots have also been able to directly integrate content from Gmail, Google Drive, and other services for some time now.

Here are a few prompt ideas that will immediately save you time and give you awesome insights:

“Analyze all incoming invoices from Gmail over the last 6 months. Show me potential savings and find 3 cheaper alternatives for each one.”
---
“Analyze email interactions with contacts from the last 6 months. Categorize relationships (e.g., clients, friends, etc.), suggest follow-up messages, and research social media profiles on LinkedIn and Instagram from the web.”
---
“Create a guide on [topic X] based on my PDFs and content in Drive. Supplement with current best practices from the web.”
---
“Summarize all email threads and chat messages related to project Y. What are the open to-dos and next steps?”
---
“Research all AI-related news from the last 7 days in Gmail and give me a ‘smart brevity’ summary with links.”
---
“Create a knowledge base on [topic/process]: all relevant how-to docs, email best practices, chat discussions with solutions, and external best practices.”
---
“Scan all relevant docs, emails, and chat threads on [topic/client] from the last 6 months. Summarize key insights, open tasks, stakeholder updates, and risks. Generate an agenda, personalized questions, and follow-up template.”
---
“Act as Chief Strategy Officer & Life Coach. Analyze my last 5 years (Drive, Gmail, chats). Identify core topics, top projects, patterns, strengths, weaknesses. Use this to form the one lever path for maximum impact & fulfillment – with a 12-month plan, milestones & risk hedging.“

Bonus tip: Export the result to Google Docs and have the report read aloud to you using the new Audio Overview feature.

Our take: Thanks to Deep Research, you'll have time for deep work again!

We think it's super handy that Google has radically expanded the Deep Research feature. Anyone who juggles Docs, Sheets, and emails on a daily basis will love it. It finally turns your data into an asset instead of a data graveyard—and you can combine your favorite prompts with Workspace data for even more specific analyses.

🔥 Better than GPT and Claude—and even open source!

Things are heating up. At the beginning of the year, we had the DeepSeek moment, which wiped out billions in market value overnight and really fueled the race between the US and China. With Kimi K2 Thinking, there is now a new model from Moonshot (backed by Alibaba) that can compete with the best top models from the US.

Kimi K2 Thinking is extremely intelligent and is now one of the top models!

In important benchmarks that "measure real intelligence and agentic behavior," the model performs well. It is the first open-source model to beat closed-source competitors such as GPT and Claude in agentic benchmarks!

What makes Kimi K2 different, and how can you test it?

We don't necessarily jump on every new model because most of us primarily use ChatGPT anyway. And even if such benchmark results sound impressive, they don't say anything about whether this is the right model for your specific use case. But we find a few things very exciting.

Forget "prompt in, response out." Kimi K2 works like an assistant who thinks, researches, plans, checks, and adapts. And it does all this without you having to specify every step.

The easiest way to try it out for free is on the Kimi platform or at OpenRouter. It's perfect for checking whether the model is suitable for your use cases. For API use, go directly to Moonshot AI, OpenRouter, or for those with their own infrastructure, download it from Huggingface.

The model thinks in terms of workflows:

• Plans up to 300 tasks independently in advance

• Calls up tools, writes code, analyses data

• Corrects itself when something does not work

• Keeps track of everything even during long tasks thanks to a 256k context window

The model is said to be particularly good at creative writing. This is an area where many other models do not always perform well...

Our take: Not just for nerds, with potential for independent agents

Kimi K2 is more than just another AI model. It shows how fast the market is moving. And that the best tools don't always have to come from Silicon Valley!

If you're already thinking about AI-based workflows and automation, you should try Kimi in your tech stack, especially if you:

• Want to automate complex research tasks

• Working with large documents or large amounts of data

• Want to optimize API costs (significantly cheaper than GPT-5 and Claude)

• Looking for an open-source alternative to the big players

The Agentic features open up truly practical use cases, from social media content creation and automated research tasks to serving as a "productive colleague" in the team for routine jobs. And all without breaking your budget!

You made it to the end—thanks for reading! We’ll be back soon with even more updates.

Reto & Fabian from the AInauts

P.S.: Follow us on social media - it motivates us to keep going 😁!
X, LinkedIn, Facebook, Insta, YouTube, TikTok

Your feedback means the world to us. We read every comment and message—just hit reply and tell us what you think!