AI-Powered Filmmaking with Veo 3

Google has launched Flow, a new AI filmmaking tool built for creatives. Designed around Google’s most advanced models—Veo, Imagen, and Gemini—Flow enables storytellers to create high-quality cinematic clips with ease.

Key Features:

• Camera Controls: Fine-tune motion, angles, and perspectives

• Scenebuilder: Extend and edit shots with seamless transitions

• Asset Management: Organize ingredients and prompts efficiently

• Flow TV: Explore community-generated content and learn from shared prompts

Now Available
Flow is the evolution of VideoFX and is available in the U.S. for Google AI Pro and Ultra subscribers. The Ultra plan includes early access to Veo 3 with native audio generation.

OpenAI’s Codex is Like Having a Personal Software Engineer

OpenAI has launched Codex, an advanced AI coding agent built into ChatGPT, offering developers an automated assistant that writes, fixes, tests, and even drafts pull requests based on plain-language prompts.

What sets it apart:

• Uses a custom codex-1 model, supporting Python, JavaScript, TypeScript, and more

• Reads your entire codebase and works autonomously in a secure, cloud-based sandbox

• Executes tasks like "Add a login page in React" with real-time logs and step-by-step reasoning

• Customizable with an AGENTS.md file for coding standards and team preferences

Currently available for ChatGPT Pro, Team, and Enterprise users, with Plus and Edu support on the way. Though not a replacement for developers, Codex significantly accelerates workflows and reduces repetitive coding.

Safety first:
Codex operates in a sealed environment without internet access and includes protections against malicious code generation. Every action is logged for full transparency.

Synthetic Data: The Missing Link for Enterprise AI Agents

According to Alexandra Ebert of MOSTLY AI, synthetic data is the key to unlocking the full potential of AI agents in enterprise settings. While AI agents promise to boost productivity and transform business operations, privacy regulations like GDPR prevent companies from using sensitive internal data to train these tools effectively.

The solution lies in enterprise-grade synthetic data—anonymized replicas of real datasets that retain analytical value without compromising privacy. This allows companies to safely train AI agents on proprietary data, enabling truly personalized and powerful applications in sectors like healthcare and finance.

As Ebert argues, the future of AI agents depends not just on smarter models, but on smarter data—synthetic data that balances innovation with compliance.

You can now play a real-time AI-rendered Quake II in your browser

Microsoft has unveiled WHAMM, a generative AI model that can render and respond to real-time gameplay—demonstrated through a playable browser version of Quake II. WHAMM (World and Human Action MaskGIT Model) is the successor to WHAM-1.6B, now using a parallel token generation method for faster visuals.

Though the frame rate remains low, the focus is on showcasing the model’s real-time AI capabilities, not on gameplay quality. Trained on just over a week’s worth of data, WHAMM reduces training time dramatically and delivers higher resolution than its predecessor.

The interactive demo is now available via Copilot Labs.

At Amazon, AI Push Makes Coding Feel Like Factory Work

As Amazon ramps up its use of generative AI, some software engineers say their roles are shifting from creative problem-solving to fast-paced, repetitive work—comparing the change to warehouse automation. Developers report tighter deadlines, less autonomy, and growing pressure to use AI tools that speed up coding, testing, and documentation tasks.

CEO Andy Jassy has emphasized AI’s role in boosting productivity and cost efficiency. While some engineers welcome the automation of tedious tasks, others worry about diminished job satisfaction, skill development, and career progression—especially among junior developers.

Amazon maintains that AI is meant to augment, not replace, human talent. Still, many workers say they now feel like bystanders in their own jobs, with AI doing more of the writing and them more of the reviewing.

The company’s engineers have begun discussing these changes more openly, raising concerns not just about AI’s impact on work quality, but also on long-term career paths and well-being.

Did you enjoy this newsletter?
Feel free to share it with a friend.

See you in the next one.