Welcome To The Future

We rarely dedicate the intro to a single story. This week we do.

The GPT-5 launch showed how hype and value follow different curves.

Expectations were set near superintelligence. The rollout was uneven. Many felt underwhelmed. Shift the lens to operator metrics like cost, latency, reliability, and integration, and the picture changes.

GPT-5 raises the floor more than the ceiling. Hallucinations are about half of the prior flagship. Throughput is faster. Accuracy is tighter. That means less fact-checking, higher first-pass usability, and more time for research and analysis.

For routine marketing work - briefs, channel variants, outlines, insight pulls - it hits what most teams want: speed with consistent quality. Turn on Thinking Mode. Always turn on Thinking Mode. Always (!!) turn on Thinking Mode!

Your prompting will need small tweaks. GPT-4o tolerated incomplete “vibe” prompts. GPT-5 behaves more like Gemini. It thrives on complete context: clear guidelines, defined deliverables, solid inputs, and unambiguous asks.

In our tests across content generation, summarization, categorization, and reporting, GPT-5 consistently outperformed earlier models when prompted correctly.

And now, let’s zoom out.

The lack of a dramatic AGI leap is a sideshow to the practical gains. If in 2020 we told you that by 2025 we would have a system that drafts better text than most people, brainstorms on demand, produces reports in seconds, and spots patterns in data at analyst-team scale, you would have called it sci-fi.

Welcome to the future.

A quick word from this week’s sponsor:

Limitless is your new superpower - an AI-powered pendant that captures and remembers every conversation, insight, and idea you encounter throughout your day.

Built for tech leaders who need clarity without the clutter, Limitless automatically transcribes and summarizes meetings, identifies speakers, and delivers actionable notes right to your fingertips. It’s securely encrypted, incredibly intuitive, and endlessly efficient.

Order now and reclaim your mental bandwidth today.

Discover Limitless

On the Q2 results call, Amazon CEO Andy Jassy signaled that Alexa+ will carry ads directly inside conversational responses. This reframes Alexa+ from a utility into a monetized media surface where brand suggestions, shoppable replies, and sponsored actions can appear mid-dialogue.

Days later, Elon Musk said X will introduce ads in Grok’s answers, pitching it to marketers as a way to fund XAI’s GPU bills and revive X’s ad business. The plan: contextual, in-thread paid mentions that sit alongside organic responses.

Together, these moves formalize a new paid channel: “answer ads” embedded in assistant outputs. Expect rapid experimentation and a lot of policy questions.

Budgets will start shifting from keyword auctions toward intent auctions, where the model infers need and offers a brand inline. Creative becomes micro-conversational (10–30 word utterances that read naturally aloud), while measurement moves to session-level IDs and post-exposure lift.

Shopify rolled out Global Catalog, Universal Cart, and Checkout Kit. Tools that let apps and AI agents search across product catalogs, add items from multiple merchants to a single cart, and check out without human intervention.

It’s a clear bet that agents will be a primary shopping channel and need first-class rails to browse, compare, and transact across millions of stores.

CEO Tobi Lütke didn’t mince words on X: “Agents will become a common way people shop,” positioning these tools as the backbone for agentic commerce on Shopify.

The implication is big. If your catalog isn’t machine-readable and your policies aren’t exposed as structured data, an agent simply won’t choose you.

Merchandising now has a machine customer. Ranking isn’t just about human UX—it’s about attribute completeness, reliable inventory signals, clear returns, and latency. This favors brands with disciplined PIM/MDM and penalizes “messy middle” catalogs.

It also opens new promo mechanics (agent-readable bundles, auto-applied offers) and forces you to think about post-purchase APIs (status, cancellations, returns) as conversion levers.

ElevenLabs rolled out Eleven Music, a generator for full tracks (including vocals and voiceovers) with licensing that covers most commercial use (ads, podcasts, and social video).

Because ElevenLabs’ core competency is high-quality voice synthesis, the new tool is designed to play nicely with narration: you can generate a music bed and layer crisp, natural-sounding voiceovers on top without the usual muddiness or clash in the mix.

Hear it for yourself in this mini-commercial we’ve prepared for the AI Ready CMO Community:

AI Voice Generator & Text to Speech

The company has also moved to pre-empt rights concerns by striking licensing deals (e.g., Merlin, Kobalt) and publishing more explicit guidance on allowed uses. That positions Eleven Music as a safer default for teams that want speed and scale without stepping into legal gray zones.

Marketing workflows often hinge on tight, on-brand narration over a sound bed—explainer videos, UGC-style spots, product demos, paid social variants. ElevenLabs’ background in voice generation means voiceovers sit cleanly over its music outputs, reducing re-takes, manual EQ, and external post.

The practical upshot: faster creative cycles, easier localization (swap script + voice, keep the bed), and a viable way to test dozens of sonic variants across markets without blowing up production budgets.

Alibaba’s Qwen-Image debuted with unusually precise, multilingual text rendering (e.g., Chinese + English) alongside strong image generation and editing—historically a weak spot for most text-to-image models. Try it here (turn on Image Generation).

Accurate native text on images enables practical marketing tasks—localized posters, packaging mockups, or social carousels—without the awkward letterforms you’ve come to expect from generalist models. Early hands-on tests and coverage reinforce the text-quality claims.

This meaningfully reduces the manual retouching tax (especially on multilingual campaigns) and speeds concept-to-mockup cycles.

It also creates a viable open-weight option for teams who need on-premise control over brand and sensitive assets.

Duolingo faced criticism for adopting an “AI-first” approach, which slowed hiring and relied heavily on automation for content. The outrage didn’t dent the numbers. Revenue and profitability beat expectations, daily actives continued to rise, and guidance increased (as well as the stock, by 30%).

Another example: Airbnb rolled out an AI customer-service agent in the U.S. and saw a 15% drop in cases needing a human. CEO Brian Chesky was clear about the scope: helpful for support, not a replacement for search at the top of the funnel (at least not yet).

This is the operating model to copy: route the repetitive majority to an agent, escalate edge cases to people, and explain the setup to customers in plain language. You protect CSAT, lower costs, and free teams for higher-value work.

The PR risk is manageable if you share outcomes and keep a human override in easy reach.

Small Headlines. Big Shifts.

» Claude Opus 4.1

Anthropic released a new version of the most advanced Claude Opus model. This release mainly focused on coding skills, the area where Anthropic already has a substantial lead over OpenAI.

» Google's Genie 3 Generates Interactive Worlds

Google has unveiled Genie 3, a model that can generate entire interactive, game-like worlds from a single prompt. While still in the early stages, this technology points to a future of fully immersive brand experiences and virtual showrooms created on demand, moving beyond static content to interactive storytelling.

» The AI Social Feed

Character.AI has launched the world's first AI-native social feed, allowing users to share and interact with AI-generated content. This points to a future where AI is not just a tool for creation, but also a medium for social interaction.

» OpenAI's Open Source Move

OpenAI has released gpt-oss, its first open-source model in years. The initial feedback has been mostly negative. The model struggles with real-world tasks and seems to be “benchmaxed” (trained to pass benchmarks with high scores). It is also tough to fine-tune.

» A Warning for Gen Z

Bill Gates has cautioned that entry-level white-collar jobs are at risk from AI, with many in Gen Z now gravitating towards skilled trades.

This week’s updates show the same pattern. The winners aren’t chasing hype, they’re redesigning how work gets done.

GPT-5, Alexa ads, Shopify’s agent rails, AI-first ops at Duolingo and Airbnb… Different stories, same lesson: the advantage goes to leaders who turn AI from a tool into an operating system.

Be one of them.

— Peter & Torsten

» PS: Our community for AI-Ready CMOs is now open for applications. We still have 19 places left for Founding Members.

Use your unique referral code to unlock even further discounts: