A Starbucks run by 100 AI-powered Robots

PLUS: Amazon's AI 'private investigator'

Welcome, AI enthusiasts.

The robot takeover has begun — and at Naver’s in-office Starbucks, the revolution is happening one cup of coffee at a time.

The company’s robot fleet navigates 36 floors to deliver orders, while providing a glimpse at the future of the service industry in the process. Let’s explore…

In today’s AI rundown:

  • Robots serve up coffee at Starbucks

  • Amazon’s AI ‘private investigator’

  • Underlord: your powerful but subordinate AI video intern

  • Microsoft’s AI weather forecasting model

  • 5 new AI tools & 4 new AI jobs

  • More AI & tech news

Read time: 4 minutes

LATEST DEVELOPMENTS

NAVER

Image source: Naver Labs

The Rundown: South Korean search giant Naver shared a video of the company’s in-office autonomous Starbucks location — with 100 robots delivering coffee and other items throughout the building.

The details:

  • Naver's autonomous wheeled "Rookie" robots navigate the building's 36 floors to bring packages, coffee, and lunch to employees.

  • Rookie bots are assisted by dual-armed "Ambidex" robots, which are lightweight and dexterous for safer human interactions.

  • Both robots connect to the company’s ARC system, which enables control of navigation, planning, and processing for the entire fleet through cloud computing.

  • Naver also developed RoboPort, a dedicated elevator system that allows robots to move quickly and efficiently between floors.

Why it matters: Naver’s putting its robots to work in the ‘real world’ — and while the location is just a testing ground for now, it’s also a glimpse into how service robotics will be integrated into the mainstream sooner than many expect.

TOGETHER WITH OCTOAI

The Rundown: Selecting the right LLM is crucial for your app's success. Join OctoAI's experts for a live discussion on how to avoid "model fatigue" and pick the perfect open-source model for your project.

You’ll learn how to:

  • Narrow down the vast universe of models to a shortlist tailored to your needs

  • Decide between checkpoints, LoRAs, and Community tunes for your specific use case

  • Evaluate model success across prompts, data, training, and tuning

Sign up for the discussion now — everyone who registers receives the recording and deck, even if you can’t attend live!

AMAZON

Image source: Amazon

The Rundown: Amazon just unveiled Project P.I., an AI system that scans products in the company’s fulfillment centers to detect damaged or incorrect items before they ship to reduce returns and waste.

The details:

  • Project P.I. uses AI and computer vision to spot defects like damaged products as well as wrong colors/sizes as items pass through imaging tunnels.

  • The system is already in place across the company’s North American fulfillment centers, with plans to expand globally throughout the year.

  • Amazon also utilizes a multimodal LLM to investigate issues further, combining customer feedback with Project P.I. images to identify the source problem.

Why it matters: Amazon’s no stranger to adopting AI across its operations, and Project P.I. is yet another innovative approach to supercharge efficiency. Soon, the company will likely combine these talents with advanced warehouse robots — taking humans out of the fulfillment process altogether.

AI TRAINING

The Rundown: Descript, an AI-powered video-audio editor, just launched Underlord, with AI editing tools built into every step of your workflow.

Step-by-step:

  1. Sign up for Descript. Download it or use it on the web for free.

  2. Upload a file or record a video.

  3. Ask Underlord to remove filler words or retakes, create clips, or do any editing tasks you hate most.

  4. Explore all the other features that Underlord has to offer with its easy-to-use interface!

PRESENTED BY TABNINE

The Rundown: Tabnine's AI code assistant adapts to your team's unique coding style and requirements, offering three levels of progressive personalization for a seamless, custom-fit experience.

With Tabnine, you can:

  • Get accurate, relevant results with context awareness inside the IDE

  • Connect to your codebase, requirements, and docs to enhance performance

  • Fine-tune Tabnine’s AI models for ultimate customization

Get started today with an AI code assistant that feels like a fully onboarded member of your engineering team.

AI RESEARCH

Image source: Microsoft

The Rundown: Microsoft researchers just introduced Aurora, a new AI foundation model trained on over 1M hours of weather and climate data that can generate accurate weather forecasting insights.

The details:

  • Aurora produces accurate forecasts across a variety of weather scenarios, including extreme events or areas with limited data.

  • The 1.3B parameter model can generate a 5-day global air pollution prediction in under 60 seconds.

  • Aurora also produces 10-day global forecasts at high resolutions, beating both the top models and specialized AI weather systems.

Why it matters: Aurora’s success could usher in a huge shift in the way we do weather forecasting. With the recent progress of AI models, combined with massive amounts of atmospheric data — the day that the weatherman is truly right 100% of the time might be coming sooner than most people probably think.

NEW TOOLS & JOBS

  • 🎬 ToonCrafter - Create animations from still-frame images

  • 🔊 ElevenLabs Text to Sound Effects - Generate any sound from a prompt

  • 👟 Doly - Capture products with your phone and create 3D videos instantly

  • 🗣️ Speller AI - Copilot to enhance your speaking skills

  • 🏦 Keeper - AI-powered tax filing for effortless returns

  • 🎨 Adept AI - Product Designer

  • ⚙️ Luma AI - Senior Distributed Systems Engineer

  • 🚗 Scale AI - Account Executive, Automotive

  • 🧑‍💻 Lambda - Technical Program Manager

QUICK HITS

Intel introduced a series of new AI advancements at Computex, including new Xeon 6 processors, Gaudi accelerators, and a new Lunar Lake architecture to accelerate computing performance.

Stability AI announced the open release of Stable Diffusion 3-Medium is set for June 12, with the 2B parameter model coming in at a fraction of the full SD3 that is still awaiting model weight release.

AMD unveiled its next-gen MI325x AI chips at Computex, announcing a similar one-year release cadence that will feature the MI350 in 2025 and MI400 series in 2026.

Microsoft announced a $3.2B investment into Sweden’s AI and cloud infrastructure, aiming to train 250,000 workers by 2027 and boost capacity at data centers.

Google published a blog addressing the company’s recent AI Overview issues, highlighting new detection mechanisms and ongoing accuracy efforts.

Adobe is facing backlash after the estate of late artist Ansel Adams called out the company for including AI-generated images for sale in Adobe Stock labeled with “Ansel Adams-style”, which violates the platform’s ToS.

Zoom founder Eric Yuan said in an interview with The Verge that ‘digital twin’ AI avatars will eventually attend Zoom meetings and make decisions on user’s behalf.

THAT’S A WRAP

SPONSOR US

Get your product in front of over 600k+ AI enthusiasts

Our newsletter is read by thousands of tech professionals, investors, engineers, managers, and business owners around the world. Get in touch today.

FEEDBACK

How would you rate today's newsletter?

Vote below to help us improve the newsletter for you.

Login or Subscribe to participate in polls.

If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.

Join the conversation

or to participate.