TLDR 2025-09-29

New research from Contentful surveyed 425 marketing decision makers (including 103 CMOs) to uncover the reality behind AI adoption in marketing.

The findings reveal a massive "optimism-execution gap" that's fast becoming marketing's fault line. While nearly every CMO claims to prioritize AI, most companies still treat it as an experiment rather than infrastructure. Learn what separates the winners from the non-starters:

• C-suite perspectives on $500K+ investments reshaping marketing ops
• Workflow insights from top performers
• Budgeting frameworks for weighing speed/quality/personalization

→ Get the report ↗️

→ Tune into the on-demand SEO to GEO webinar featuring leaders from Contentful, Vercel, and Apply Digital

Apple is currently using an internal ChatGPT-like iPhone app code-named Veritas to help test and prepare for a Siri overhaul planned for next year. The app is being used to quickly evaluate new features, including the ability to search through personal data and perform in-app actions. The internal tool marks a new phase in Apple's preparations for the Siri revamp, which could help reshape perceptions of its AI efforts.

Meta's chief technology officer, Andrew Bosworth, recently said in a meeting that he wants to focus on licensing Meta's software to other manufacturers instead of being a hardware manufacturer. Meta's new Superintelligence AI lab is working with the company's robotics group to build a 'world model' that can do software simulations to animate dexterous movements. Bosworth says that robotics will be Meta's next 'AR-sized' bet. The company has so far spent over $100 billion in cumulative investment into its augmented and virtual reality portfolio.

SpaceX currently launches and lands its Falcon family of rockets every two days on average without interrupting the operations of other launch providers. Launch sites of the future will need to be fully operationalized like an airport, with multiple launches a day from multiple providers. SpaceX is working with federal regulators, the federal ranges, and industry partners to realize this vision. Its goals will advance the entire launch industry's ability to fly rapidly and reliably, ideally from multiple locations to increase resilience.

Current projects developing humanoid robots will fail at teaching them dexterity because, in order to succeed, they need both the right data and for the robots to learn the right thing. Human dexterity relies on a rich sense of touch, and it often involves elbows, the front of the body, legs, and feet. Just using visual data is not enough. In the future, there will likely be a lot of robots with different forms for different specialized jobs that humans can do. They won't look human, but they will probably all still be called humanoid robots.

If slow QA processes bottleneck you or your software engineering team and you're releasing slower because of it — you need to check out QA Wolf.

They get engineering teams to 80% automated end-to-end test coverage and help them ship 5x faster by reducing QA cycles from hours to minutes.

QA Wolf takes testing off your plate. They can get you:

• Unlimited parallel test runs
• 24-hour maintenance and on-demand test creation
• Human-verified bug reports sent directly to your team
• Zero flakes, guaranteed

The result? Drata's team of 80+ engineers achieved 4x more test cases and 86% faster QA cycles.

No flakes, no delays, just better QA — that's QA Wolf.

Schedule a demo to learn more

Taste is the ability to adopt the set of engineering values that fit the current project. It's different from skill - it's possible to be technically strong but have bad taste or be technically weak with good taste. Almost every decision in software engineering is a tradeoff. Often, you have to make tradeoffs between engineering values, such as resiliency, speed, readability, correctness, and flexibility, and that's where taste comes in. Bad taste is when your preferred values are not a good fit for the project.

Small Data is a movement away from Big Data toward much simpler and cheaper solutions. It is being driven by improvements in hardware and organizations that don't use that much data. Distributed systems are more overkill than ever. Developers should be considering what technologies they know well, are good enough, and are the least toilsome for the task at hand. Modern hardware will likely handle the workload just fine.

Many people look at the current state of AI, see the mistakes it makes, and then somehow jump to the conclusion that AI will never be able to do tasks at a human level. A few years ago, our current AI models were just science fiction. AI is improving at an exponential rate. It will be surprising if these improvements suddenly stop. Even a relatively conservative extrapolation of the trends suggests that by the end of 2027, models will frequently outperform experts on many tasks.

Open source has clearly won, but this wasn't a foregone conclusion 35 years ago. There were many powerful forces that wanted open source to lose. We are now at a similar juncture with social apps. Projects like Bluesky's AT Protocol are changing the relationship between the user, the developer, and the project. What open source did for code, open social does for data.

Electric is one platform to handle devices, users, procurement, and security - built for companies with 50-1000 employees. And they're giving away $200 gift for qualified prospects who take a demo. Claim the offer

TLDR AI is your daily fix of LLMs, GenAI, and deep learning goodness. Same TLDR format. Still free.

Subscribe now.

The systems have been unreliable, expensive, slow to be manufactured, or couldn't be bought in the quantity needed.

Avalanche Studios was born out of ambition, but it failed to deliver on that ambition and closed a year ago.

Accenture's main strategy is upskilling, and it will exit people so it can get more of the skills it needs.

Turning compute into a tradable commodity will be needed to fuel AI's next stage.

Figma proved design doesn't need Illustrator, Descript proved editing doesn't need Premiere, and Omarchy and Framework proved developers don't need macOS.