Good morning, {{ first_name | AI enthusiasts }}. The future of advertising just crashed the NBA Finals, with an “unhinged” commercial created in just two days airing on ABC.
Kalshi’s spot just proved that AI can go from prompt to primetime in just 48 hours (and a 95% cost reduction) — but will brands be willing to embrace the chaos, or wait for more polished creations?
P.S.: Our next live workshop is today at 4 PM EST, where you’ll learn how to prompt and get the most value out of OpenAI’s advanced o3 and o3-Pro models. RSVP here.
In today’s AI rundown:
Kalshi’s AI ad runs during NBA Finals
OpenAI, Mattel partner on AI toys
Conduct competitive analysis with Claude
ByteDance’s new video AI climbs leaderboards
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
AI RESEARCH
Image source: Kalshi
The Rundown: Prediction market Kalshi aired one of the first instances of an AI-generated commercial during Game 3 of the NBA Finals, running an “unhinged” 30-second spot created with clips created using Google's Veo 3 video model on ABC.
The details:
AI filmmaker PJ Accetturo created the ad in just 2 days, using 300-400 Veo 3 generations to create 15 clips.
He detailed his workflow in a post on X, using Gemini and ChatGPT to help with ideation, script creation, and craft prompts for each shot.
The commercial leveraged Veo 3's new speaking capabilities, though Accetturo noted challenges with unexpected subtitles and inconsistent character voices.
Accetturo estimated the cost at about 95% less than traditional production, and said that “high-dopamine Veo 3 videos will be the ad trend of 2025.”
Why it matters: This feels like a major moment for AI in advertising on a big-time stage — and while its unpolished, chaotic aesthetic won’t work for every brand, the time and cost element will be hard to resist for many advertisers. The “style” is also unlikely to be an issue for long, with video models only getting better with each new release.
TOGETHER WITH SANA
The Rundown: Sana Agents lets every team spin up intelligent, no-code AI agents in minutes — and now its new Strange Loop podcast digs into the edge of AI and human knowledge to keep leaders ahead of the curve.
In the debut episode, Sana CEO Joel Hellermark joins Wharton’s Ethan Mollick to:
Unpack myths about AGI
Discuss the coming impact of real-world automation
Break down how to scale an AI-first company
OPENAI & MATTEL
Image source: GPT 4o / The Rundown
The Rundown: OpenAI and Mattel just announced a new strategic partnership to create AI-powered toys and experiences, bringing the tech to franchise brands like Barbie, Hot Wheels, American Girl, and more.
The details:
The collaboration will integrate OpenAI's tech into Mattel's product development, with the first AI-powered product expected later this year.
The deal covers physical toys and digital experiences across Mattel's portfolio, featuring hundreds of iconic brands and game titles.
Mattel employees will also gain access to ChatGPT Enterprise to enhance creative ideation and streamline business operations across the company.
Both companies emphasized safety and age-appropriate design, with Mattel maintaining full control over its IP and final products.
Why it matters: Toys are about to get a lot more personalized and intelligent with an AI infusion, bringing new interactive capabilities like never before. But having children interact with AI in any fashion is likely to ruffle some feathers and raise questions about development, privacy, and whether the tech belongs in the playroom.
AI TRAINING
The Rundown: In this tutorial, you will learn how to use Claude's Research mode to automatically create research plans, gather data from hundreds of sources, and generate downloadable PDF reports for complete market analysis.
Step-by-step:
Head over to Claude and select the Research button
Create a comprehensive prompt: “Research the [industry] market landscape, analyze top 5 competitors, investigate pricing strategies, partnerships, and identify 3 key opportunities”
Watch Research mode automatically gather data from 50+ authoritative sources and generate a detailed report
Download your complete research as a PDF or Markdown
Pro tip: The more comprehensive your initial prompt, the better the automated research plan and final report will be.
PRESENTED BY AIRIA
The Rundown: Airia is an enterprise platform that simplifies the creation, orchestration, and deployment of AI agents—so businesses can automate workflows, integrate systems, and unlock real value from AI without the complexity.
Airia allows you to:
Seamlessly route between any model or system and connect to 50+ enterprise tools
Build AI agents fast with a no-code builder to design and launch in minutes
Enterprise-grade control to enforce policies, manage spend, and stay compliant
BYTEDANCE
Image source: ByteDance
The Rundown: ByteDance just released Seedance 1.0, a new video generation model that ranks first on benchmarks for both text-to-video and image-to-video tasks, outperforming SOTA options from Google, Kuaishou, and OpenAI.
The details:
Seedance 1.0 moves to the top of the Artificial Analysis video leaderboards, moving ahead of top models including Veo 3, Kling 2.0, and Sora.
The model generates 5-second, 1080p videos in 40 under a minute, with multi-shot storytelling, character consistency, and smooth transitions.
Bytedance also created SeedVideoBench, a benchmark that shows its model ahead of competitors in motion quality, prompt adherence, and aesthetics.
The company plans to fold Seedance into its Doubao chatbot and video platform Jimeng later this year.
Why it matters: Despite the impressiveness of Veo 3, it’s already been surpassed on the leaderboard — with Chinese labs continuing to push the frontier for AI video models. But with Google’s Veo 3 viral audio capabilities setting a new expectation for consumers, it makes it hard to go back to still “silent” models.
QUICK HITS
🌐 Dia - The Browser Company’s new AI-first web browser
🎥 Meta AI - New video editing capabilities for short-form content
🏄 Windsurf Browser - Integrated with coding assistant for in-context support
💬 Manus Chat Mode - Free, unlimited AI chat with instant answers
Bytedance researchers introduced Seaweed APT2, a new model for real-time, interactive video generations — able to stream 24 fps videos at up to 5 minutes long.
Microsoft rolled out Copilot Vision with highlights in the U.S., allowing the assistant to see users’ screens and provide in-context insights and guidance.
Google DeepMind launched Weather Lab, an interactive platform showcasing its AI-powered weather forecasts for early, accurate predictions of storm paths and intensity.
Apple is reportedly targeting Spring 2026 for its AI-powered upgrades to Siri, which would come almost two years after its introduction at WWDC 2024.
Runway released Chat Mode, a new conversational interface to create images, videos, and more using natural language.
AMD introduced its next-gen Instinct MI400 chips in a presentation alongside OpenAI CEO Sam Altman, positioning itself as a lower-cost alternative to Nvidia.
Los Alamos, Meta, and Berkeley Lab released Open Molecules 2025 with 100M+ molecular simulations for training AI for chemistry, drug discovery, and more.
COMMUNITY
Join our next workshop today at 4 PM EST with Dr. Alvaro Cintas, The Rundown’s AI professor. By the end of the workshop, you’ll know how to get significantly more value out of whichever model you use, whether it’s o3 or o3-Pro.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
That's it for today!
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
