Did Grok 4's Launch Miss the Mark?

Great product launches, especially in the consumer-focused AI space, are about storytelling and connection. They must "show, not just tell." This is where the Grok-4 presentation first faltered. Instead of leading with an inspiring, relatable demonstration of its new capabilities, the team spent the initial 20 minutes focused on performance benchmarks. While impressive to a niche group of developers, these charts are abstract and uninteresting for the broad audience a product like this needs to capture. It stands in stark contrast to OpenAI's recent launches, which captivated viewers by showing seamless, human-like conversations and real-time visual understanding. xAI had an opportunity to show something cool and instead chose to show a graph. For a broad audience tuning in to see what makes Grok 4 special, this felt like watching someone recite their SAT scores at a party.

When they finally got to the demos, the choices were puzzling. The first major showcase? Using Polymarket integration to help people bet more effectively on sports. Even as someone who appreciates prediction markets, leading with gambling feels tone-deaf, especially when you're essentially advertising AI-powered sports betting to people who may not have known it existed.

The voice capabilities demo was equally strange. Rather than showcasing natural conversation or practical use cases, we watched an AI count from one to five, followed by an uncomfortable ASMR-style interaction featuring what seemed like a sexualized British voice awkwardly singing about Diet Coke. It wasn't so bad it was funny, and it certainly wasn't good, it was just weird.

The gaming section highlighted another fundamental misunderstanding. While games created by LLMs has huge potential, general AI models aren't likely to compete with specialized tools. A dedicated gaming LLM will always appeal more to game developers than whatever X cobbles together. It's the same trap that's caught OpenAI with Sora, trying to build an encyclopedia that's also creative rarely works well.

The Enterprise Problem

What really limit’s Grok 4’s potential is the uncertainty around whether what it says will damage a company’s brand. If companies can't trust that your AI won't say something brand-damaging, they won't use it. Grok's reputation for fewer guardrails might appeal to some users, but it creates a massive blind spot in the enterprise market.

The coding capabilities felt like another missed opportunity. Rather than showing practical improvements or innovative approaches to development, we got Elon suggesting people just copy their entire codebase into the 256K context window. Something that works for one-off questions, but not serious AI-driven development

A Question of Taste

Compare this to how Apple used to handle product launches. Steve Jobs understood that you lead with the magic, not the metrics. You show people something that makes them say "wow," then worry about the technical details later. Even OpenAI's GPT-4 launch, while technical, led with clear, compelling use cases that people could immediately understand and want.

The fundamental issue isn't technical capability, it's taste. X and xAI consistently build things that feel unnecessary, driven by a warped understanding of what actually appeals to people. The dark room, the awkward demos, the focus on gambling and gaming, it all suggests a company that's lost touch with what makes products compelling.

This matters because in the AI space, technical capability is becoming table stakes. GPT-4, Claude, Gemini, they're all remarkably capable. What differentiates them is how they're positioned, presented, and integrated into people's workflows. When your launch feels like it was designed by and for people who think counting to five is impressive, you're not going to win mainstream adoption.

Grok 4 might have the technical chops to compete, but until X and XAI figure out how to present their work with genuine taste and understanding of what people actually want, they'll remain a niche player in an increasingly crowded market. In AI, as in so many other industries, how you show up matters as much as what you ship.