Good morning. Apple may flip its playbook and brand every upcoming OS by year—think iOS 26 and macOS 26—setting the stage for a unified lineup at WWDC. Interested to see what iOS 70 looks like…

— The Deep View Crew

In today’s newsletter:

• 📖 AI for Good: Tennessee school uses AI to improve reading skills

• 🤖 Telegram and xAI team up to bring Grok to a billion users

• 👾 Odyssey’s AI brings 3D worlds to life

📖 AI for Good: Tennessee school uses AI to improve reading skills

In Sumner County, just north of Nashville, Tennessee, schools are turning to AI to close achievement gaps in reading. 

With more than 30,000 students across 52 schools, the district is using AI to make English language arts instruction more responsive and individualized.

What happened: Director of Schools Scott Langford introduced Coursemojo, an AI tool that embeds short adaptive modules into ELA lessons. These modules challenge students to deepen their understanding while giving teachers real-time feedback on individual performance. The system also suggests whole-class follow-up questions based on student progress, acting as a digital co-teacher rather than a replacement.

Langford says the district’s strategy was clear from the start: build a strong instructional foundation, then find ways AI can enhance—not dictate—lesson delivery. Coursemojo fits that model, offering actionable insights without overwhelming staff or putting kids on screens all day.

Privacy concerns were addressed early, with protocols in place before any student data was shared. That preparation allowed the district to focus on outcomes rather than risk management.

Why it matters: Reading proficiency remains a challenge across the state, and teachers are drowning in data without the time to act on it. AI tools like this one help bridge that gap by making it easier for educators to spot where students need help. Rather than chasing trends, Sumner County is using AI with purpose, precision and respect for the teacher’s role in the classroom.

The Cheat Sheet To Maximizing AI For Your Small Biz

Running a small biz ain’t easy – and that’s before you even start diving into all the AI tools, solutions, and add-ons out there. Lucky for you, Salesforce has condensed everything small business owners need to know about AI into one simple, powerful guide:

Download “AI For Small Business” right here.

This free resource covers the most useful AI tools, the smartest ways to implement them, and the best practices you should be following. Learn how to improve customer service, cut costs, increase productivity, all with the help of AI. 

This guide is so useful, it probably shouldn’t be free… but hey, Salesforce loves helping small businesses win. Click here to get your free copy of “AI For Small Business”. 

🤖 Telegram and xAI team up to bring Grok to a billion users

Telegram is putting AI at the center of its platform through a new partnership with Elon Musk’s xAI. The move will give users access to Grok, a chatbot designed to boost productivity and communication inside the app.

What happened: Telegram signed a one-year deal with xAI to distribute Grok across its chat platform. In return, xAI will pay $300 million in cash and equity. Telegram will also collect 50% of all revenue from Grok subscriptions purchased through the app.

Grok will be available to more users, beyond just Telegram Premium. A video shared by CEO Pavel Durov showed Grok pinned to the top of chats, accessible through the search bar, and able to summarize messages, answer questions, suggest replies, and even create stickers. Businesses can also use Grok to help with moderation and customer interactions.

This expansion follows a limited release earlier this year, which gave premium users early access to the tool.

Why it matters: Telegram is not just adding another chatbot. It is making AI part of everyday communication for more than a billion people. 

With Grok built into chats, search, and productivity tools, users can get real-time help without switching apps. The deal also creates a new revenue stream for Telegram and pushes AI deeper into mainstream messaging.

Get AI-Ready: Announcing Monte Carlo’s Observability Agents and Unstructured Data Monitoring

AI-readiness is at the forefront of every data leader’s mind.

Monte Carlo is pioneering the roadmap for AI-ready data with their industry-first Observability Agents and support for unstructured data monitoring. 

Here’s what’s launching and why it matters:

• Observability Agents—Accelerate the detection and resolution of data quality issues by deploying AI-powered data quality monitors and identifying root causes with Monte Carlo’s observability agents.

• Unstructured Data Monitoring—Deliver observability for unstructured data assets, including documents, chat logs, and more, without writing a single line of SQL.

Learn more about the launch below or at Snowflake Summit Booth #1508 and Databricks Data+AI Summit Booth #602.

• Audi taps AI to speed up production and cut inefficiencies

• DeepSeek rolls out a major R1 upgrade as the AI arms race intensifies

• China’s answer to vibe coding

• Google Photos gets a glow-up with smarter, AI-powered editing tools

• Developer builds tool that scrapes YouTube comments and uses AI to predict where users live

• LinkedIn: AI Engineering Manager, Foundational AI Data

• Qualcomm: Artificial Intelligence for Video Compression - Research Scientist

• DoNotPay: An AI legal assistant that was created because the founder didn’t like to pay for things. Now it helps users fight parking tickets, cancel subscriptions, and navigate small claims

• Levity: Automate your freight emails and calls by pulling info from your inbox and handling follow-up tasks

• Mistral Agents API: Build AI agents with Mistral

• SchedX: Turn your website into a talking AI SDR

👾 Odyssey’s AI brings 3D worlds to life

San Francisco startup Odyssey has unveiled an AI model that generates video worlds in real time, letting users move through scenes as if inside a living video game. The company calls it "interactive video" – essentially, video you can both watch and interact with, imagined entirely by AI in real time.

How it works: The web-based demo puts viewers in a first-person perspective, using game-like WASD controls to walk around AI-generated environments. In motion, it feels like walking through a blurry version of Google Street View, with scenes that resemble real locations but appear unstable and dreamlike.

• Under the hood, the system uses a "world model" AI network trained to predict video frames based on user actions. As the user moves forward or turns, the AI paints the next frame of the scene to match the motion.

• The model streams new video frames in as little as 40 milliseconds, roughly 25 frames per second. There is no game engine or pre-rendered 3D environment – every view is generated on the fly by AI.

• Odyssey allows only a short session (about 2.5 minutes) per run in this research preview, underscoring its experimental status.

The AI-generated environments are often blurry and distorted, and they can shift in unpredictable ways as you move. Walk forward far enough or stand still too long, and you might see buildings reshape or disappear – a doorway can turn into a solid wall when approached. The system sometimes lacks consistent "collision" physics: one moment a fence might stop you, while in another instance you can walk through a wall.

The costs: Generating an interactive video feed is computationally heavy. The model currently streams up to 30 frames per second by tapping clusters of NVIDIA H100 GPUs in the cloud. Odyssey estimates a cost of $1–$2 in cloud GPU time per user per hour at present.

Real footage and big backers: One thing setting Odyssey apart is its data and pedigree. The company – founded by self-driving car pioneers Oliver Cameron and Jeff Hawke — built a custom 360-degree camera rig worn like a backpack to capture high-resolution, real-world footage for its models. By gathering its own immersive video data, Odyssey hopes to ground its world models in reality and achieve higher fidelity than systems trained purely on random web videos.

Odyssey's ambitious approach has attracted serious investors and a dash of Hollywood magic. The startup has raised $27 million to date from backers including EQT Ventures, Google's GV and Air Street Capital. Notably, Ed Catmull, legendary cofounder of Pixar and former Disney Animation president, has joined Odyssey's board of directors.

Competition context: Odyssey is entering a fast-emerging arena where startups and tech giants alike are chasing generative video and AI world-building. New York-based Runway has been a trailblazer in text-to-video generation, though its Gen-2 model creates only short clips that are often low in frame rate and quality. Another startup, Luma AI, approaches the problem from the 3D angle, training models on a massive cluster of ~3,000 NVIDIA A100 GPUs to create 3D objects from text prompts.

Potential applications: If interactive AI video matures, it could transform entertainment (dynamically generated movies where the viewer is also the player), advertising (brands inviting customers inside their commercials), training and simulation (AI-generated scenarios for corporate training or defense), education (virtual field trips to historical sites) and creative workflows (rapid prototyping for film and TV production).

Odyssey's interactive video demo offers a tantalizing peek at the future of spatial computing. It doesn't take much imagination to see how this technology could converge with AR/VR hardware in the coming years. The Holy Grail is a Holodeck – an immersive simulation that feels real. Odyssey is still far from that level of fidelity, but it represents the kind of breakthrough needed to get there.

Those H100 GPU clusters don't come cheap. As the technology advances, it could dramatically lower the cost of content production. Virtual location shoots could replace flying film crews to remote sites. And if a single artist can generate what used to require a whole effects team, production budgets could shrink, potentially shifting the labor market in creative industries.

Decades ago, Pixar's Catmull and others struggled with crude, noisy renders as they pioneered 3D animation; over time, those techniques matured into the rich visuals we take for granted. We may be witnessing a similar inflection point for AI-driven content. Odyssey's demo is a rough sketch of that future: imperfect and even absurd at times, yet also staggering in its premise.

🤔 Your thought process:

Selected Image 1 (Left):

• “In the fake image, the steam and smoke are behaving as if it's a nice, breezy day. Not like they're in a hellaciously dynamic local space. Similarly a large wildfire creates its own local ecosystem when it comes to atmospheric behavior”

• “I'm from Hawaii. I've been to Kilauea. Volcanoes are our thing here. The other image was totally AI generated.”

Selected Image 2 (Right):

• “Should have stuck with my first choice – the fake is oversaturated and cone shape seemed off. But the lava patterns in the real seemed progressively generated...”

• “I've missed 3 in a row now and I am pretty aggravated.”

  • Well, I hope you get today’s, because it’s not easy :)

💭 A poll before you go

Thanks for reading today’s edition of The Deep View!

We’ll see you in the next one.

P.S. Enjoyed reading? Take The Deep View with you on the go! We’ve got exclusive, in-depth interviews for you on The Deep View: Conversations podcast every Tuesday morning. Subscribe here!

If you want to get in front of an audience of 450,000+ developers, business leaders and tech enthusiasts, get in touch with us here.