🗞️In this edition

• Microsoft unveils first In-house AI models

• Workflow Wednesday #34 ‘Streamlining Operations’

• Experts question Google Gemini AI footprint claims

• In other AI news –

  • China plans 300,000 driverless taxis by 2030

  • Anthropic to train AI on chat transcripts

  • Baidu launches AI platform with Chinese chips

• 4 must-try AI tools

Microsoft debuts first AI models to power Copilot and consumer tools

♨️News – Microsoft has introduced its first in-house AI models, MAI-Voice-1 and MAI-1-preview. The launch shows the company’s shift toward building its own consumer-focused AI systems while expanding what Copilot can do.

👨‍💻Fast facts –

• MAI-Voice-1 generates one minute of audio in less than a second on a single GPU.

• It already powers Copilot Daily news recaps and podcast-style explainers.

• Users can try it in Copilot Labs, with options to adjust voice and style.

• MAI-1-preview was trained on about 15,000 Nvidia H100 GPUs and handles text-based tasks.

• Microsoft is testing MAI-1-preview on the benchmarking platform LMArena.

🤓OpenTools POV –  By developing its own models, Microsoft is taking a step toward reducing reliance on OpenAI while tailoring AI for everyday use. This could mark the beginning of a broader shift toward specialized, consumer-driven AI experiences.

Read More

This Week in Workflow Wednesday #34: Streamlining Operations

 Workflow 1: Create Internal SOPs in Seconds (Using Guidde)
Every week, someone asks the same question: “How do I update this in Beehiv?” “Where do I click in Asana?” Instead of repeating yourself, capture the process once and scale your answer with Guidde.

Step 1: Install Guidde
Add the free Guidde Chrome Extension.

Step 2: Start Recording a Process
Open the tool you want to document, hit Record in Guidde, and walk through your workflow once…We break it all down in this week’s issue, along with 2 other AI workflows.

Google’s Gemini study downplays AI water and energy impact, experts warn

☕News –  Google says its Gemini AI assistant uses only a fraction of the water and electricity that past studies have suggested. But researchers warn the company’s new study paints an incomplete picture by leaving out key measures of energy and water use.

👨‍💻Fast facts –

• A Gemini text prompt uses about 0.26 milliliters of water and 0.24 watt-hours of electricity.

• Google equates this to roughly five drops of water and less energy than nine seconds of TV.

• Experts say the study omits indirect water use tied to electricity production.

• Researchers also note Google excluded “location-based” carbon emissions, which show local impact.

• Google claims major efficiency gains, with electricity use per prompt reduced 33 times since 2024.

🤓OpenTools POV –  Google’s numbers suggest real improvements in AI efficiency, but the gaps make it hard to judge the true cost. Until companies share more complete data, the environmental trade-offs of AI will remain murky. 

Read More

📢 What else is happening?

• China to have 300,000 driverless taxis in 4 top-tier cities by 2030: UBS // ‘Given the fast development of technologies and the increasingly expensive labour cost, autonomous vehicles’ will drive productivity: UBS

• Anthropic will start training its AI models on chat transcripts // You can choose to opt out

• Baidu unveils AI computing platform powered by Chinese chips to push a domestic tech stack // The Baige 5.0 infrastructure platform aims to raise the efficiency of DeepSeek’s open-source AI models

👩🏼‍🚒Discover mind-blowing AI tools

1. ComfyUI Web - Allows users to generate realistic and high-resolution images from text descriptions (Free)

2. Banner GPT - An AI tool that generates banner images for blog posts (Free)

3. Jobscan - A job search tool that helps job seekers optimize their resumes for Applicant Tracking Systems (ATS) (Free, $29.99/month) 

4. Civitai - An online platform that makes it easy for people to share and discover resources for creating AI art (Free)