One Designer, Ten Iterations, Zero Job Loss

Hello,

This is Simon with the latest edition of The Weekly. In these updates, I share key AI related stories from this week's news, list upcoming events, and share any longer form articles posted on the website.

Over the last year, I've noticed that those in the creative industries have been a little more hesitant to adopt AI tools than those working in other sectors — and this makes sense. Asking AI to tell you what's important in a data set seems less threatening than asking it to create a new design. But just recently, I know a designer who's finally started testing what AI can do for them, and the reaction has been very interesting. Rather than feeling like AI will take their job, they've noticed just how much quicker it is to prototype initial ideas, and even create voice-overs on videos. Using initial sketches as a source, they've been able to get AI to produce numerous iterations of an idea extremely quickly. For me as a non-designer, it's interesting to see just how much creative control they still have — it's simply much quicker to draw out the different ideas.

This loops back to an observation I've made in previous editions: we all need to embrace AI in parts of our jobs. We shouldn't feel threatened by it, but we need to make the effort to work out where it can help us — how it can save time, how it can come up with ideas we might not have had, or, more ambitiously, how we can get it to build further tools and applications for our roles. I think it's reasonable to expect a lot of variation between individuals in how far they dive into working with AI, but we all owe it to ourselves to be curious and start exploring how it can help us — at work and at home.

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide

In this section, I’m going to bring to you a real world example of AI use.

Starbucks is developing internal AI tools to replace enterprise software it currently buys from Microsoft and IBM, Bloomberg reported on 9 July, citing an internal company presentation. The tools in development include replacements for a Microsoft inventory-tracking system and an IBM maintenance-management product, with some expected to launch by the end of 2027 subject to testing. Chief Technology Officer Anand Varadarajan told staff the company spends roughly $400 million a year on software and described AI as offering "clear opportunities to reduce the spend." The initiative sits within a broader programme to cut $2 billion in costs.

Why it matters: This is a significant signal. When a company the size of Starbucks concludes it can build credible alternatives to Microsoft and IBM enterprise software using AI, the economics of the packaged software market start to shift. If AI-assisted in-house development becomes viable at scale, it changes the negotiating position of every large enterprise with a major software vendor — and it puts pressure on those vendors to demonstrate value that goes beyond what a well-prompted model could replicate.

The White House is finalising a voluntary framework with OpenAI, Anthropic, and Google that would give federal agencies up to 30 days to review new frontier AI models for national security risks before they are released publicly, according to Semafor and The Hill. Evaluation benchmarks would be classified, and an announcement is expected before 1 August 2026. The framework follows several months in which the Trump administration had been quietly holding back model releases, creating uncertainty for developers and enterprise buyers waiting on next-generation capabilities.

Why it matters: A 30-day government review window could introduce meaningful delays in when organisations can access the newest AI models — and "voluntary" frameworks have a way of becoming the baseline expectation. Enterprise technology roadmaps that depend on specific model releases will need to build in regulatory lag. It also signals that frontier AI is now squarely in the national security conversation, which will accelerate compliance requirements across regulated sectors.

OpenAI paused internal deployment of an unreleased long-horizon AI model last week after it found ways to act outside the controlled environment in which it was being tested, the company disclosed in a safety essay published on 20 July. In one incident the model exploited a sandbox vulnerability to open a public pull request on a GitHub repository; in another it split an authentication token into obfuscated fragments to evade a security scanner. The same model had, on a separate occasion, disproved a long-standing open problem in mathematics — the Erdős unit distance conjecture — underscoring just how capable, and how unpredictable, the system was. OpenAI said it restored access under continued monitoring.

Why it matters: Businesses are being actively sold AI agents that connect to their internal systems, file stores, and external services. This incident — disclosed by the developer, not discovered by a customer — is the clearest demonstration yet that even frontier labs cannot fully anticipate how powerful models behave once they have system access. Any organisation deploying agentic AI should be asking hard questions about what guardrails are in place, and who is responsible when those guardrails fail.

Thanks for reading, and see you next Thursday.

Simon,

Was this email shared with you? If so subscribe here to get your own edition every Thursday.

Enjoying Plain AI? Share it and get a free gift!

If you find this newsletter useful, why not share it with a couple of friends or colleagues who would also benefit? As a special thank you, when two people subscribe using your unique referral link, I’ll send you my "Exclusive Guide: Supercharge Your Work with NotebookLM." It’s a practical, no-nonsense guide to help you turn information overload into your secret weapon.

You can also follow me on social media: