Hey friend,

Here’s what we’re covering this week:

• Meta Unveils Llama 4: Advanced Multimodal AI Models

• Runway Releases Gen-4

• Midjourney Launches V7 AI Image Generator

• Automate Your Meetings With AI (N8N + Fireflies)

Let's dive in.

AI News

Here’s my pick of AI news this week:

1. Meta Unveils Llama 4: Advanced Multimodal AI Models

Meta has released Llama 4, a family of AI models. Scout and Maverick bring powerful, open-weight AI that natively handles text and images. They're also giving a sneak peek at Llama 4 Behemoth, a massive model already flexing its muscles against top competitors while still in training.

• These are Meta's first models built with a Mixture-of-Experts (MoE) architecture right from the start, making Llama 4 Scout and Maverick efficient for multimodal tasks.

• It's a big deal for open AI, offering serious competition to closed systems and bringing sophisticated skills – like understanding multiple images or massive text inputs – to more developers.

• Llama 4 Scout boasts a massive 10 million token context window, great for tackling giant documents or code, while Maverick competes with models like GPT-4o and Gemini 2.0 Flash on important AI tests.

2. Runway Releases Gen-4

Runway launched Gen-4, its advanced AI model that creates stunningly consistent video and images using a single reference picture. It marks a significant step towards production-ready AI content with realistic physics and control. I've been enjoying creating with it this week.

• It's Runway's latest generative AI, capable of producing video and images where characters, objects, and styles stay remarkably coherent across scenes, all driven by one reference image and your text descriptions.

• This game-changer is because it tackles AI video's most significant hurdles – consistency and control – making it practical for professional use in films and ads by simulating physics and maintaining visual integrity without needing extra fine-tuning.

• Gen-4 introduces 'World Consistency' to keep the look and feel uniform even from new angles. It simulates real-world physics for believable movement, opening doors for everything from consistent characters in series to AI-generated product shots.

3. Midjourney Launches V7 AI Image Generator

Fresh after OpenAI made considerable updates to its image generation, Midjourney launched V7, its first major AI image model update in nearly a year. It focuses on higher quality, better prompt understanding, and personalised results.

• V7 is Midjourney's latest AI image generator, built on an entirely new architecture designed to produce visuals with significantly improved quality, textures, and coherence, especially for details like hands and bodies.

• It stands out because it's the first version with personalisation enabled by default. You rate images to teach your style, making the AI adapt to your visual taste and signalling a move toward more user-centric creative tools.

• V7 is much better at processing text prompts accurately and even includes a faster, lower-cost 'Draft Mode' for when you need quick generations. However, some features like upscaling aren't available just yet.

OpenAI 4o image generation is tough competition for Midjourney.

Tools I’ve Been Exploring

Pydantic AI: PydanticAI is a Python agent framework designed to make it less painful to build production-grade applications with Generative AI. If you want to build AI agents in Python this is a great choice.

Vapi: Vapi helps developers build, test, and deploy voice agents at scale.

Automate Your Meetings With AI (N8N + Fireflies)

Last week I published a tutorial that shows how to use Fireflies AI with N8N.

Coaches, managers, and modern workers spend insane amounts of time in meetings. CEOs spend an amazing 72% of their work time in meetings. In this video, I explain how to build meeting automations so you can get transcripts, action items, and insights from your meetings stored in one central database.

Your Opinion Matters

What did you think of today’s email? Your feedback helps me create better emails for you!

Got more feedback or want me to cover a specific topic? Reply to this email and let me know.

Owain