Welcome to this edition of Loop!

To kick off your week, I’ve rounded-up the most important technology and AI updates that you should know about.

‏‏‎ ‎ HIGHLIGHTS ‏‏‎ ‎

Let's jump in!

Most AI agents can run a task. The problem is everything around it: setup, memory, context, cost, and figuring out what actually happened.

SureThing turns useful AI skills into autonomous agents with business context, persistent memory, cost-aware model selection, and a live dashboard. Paste a link, assign the work, and your agent reports back like a human teammate: what it did, what it cost, what needs your decision, and what happens next.

Built for founders, operators, and marketers who want AI to ship work, not become another tool to babysit.

Create Your Agent

1. Anthropic launches 10 AI agents for finance

We start this week with Anthropic, who have released ten AI agents for the financial sector. Each one targets a different task - whether it’s building pitchbooks, screening KYC files, or running the month-end close.

Anthropic's pre-built agents are available as plugins, so that they can be easily used within Cowork or Claude Code. You can also use them if your company has an account on Anthropic's Managed Agents platform.

This should allow finance firms to more quickly adopt and deploy their own agents, rather than having to build these agents themselves over several months.

Claude has also been added to Microsoft Excel, PowerPoint, Word, and Outlook through new add-ins. These let Claude carry context between apps, so an analyst can start their own finance model in Excel and then create a pitchbook in PowerPoint - without having to re-explain everything to the AI.

It's a smart bet from Anthropic, who have recently overtaken OpenAI and are now the leading AI company for enterprises. These templates should allow finance firms to more easily adopt the technology into their workflows, but they'll still need to tweak them for their specific use cases.

If you want to see the full list of templates from Anthropic, I've included a link below.

2. Musk wants to spend $119 billion on his own chip factory

Elon Musk has filed plans to spend up to $119 billion on a new semiconductor factory in Texas, with an initial investment of $55 billion.

The project is a joint effort between SpaceX, Tesla, xAI, and Intel - aiming to produce advanced chips for future AI servers, autonomous vehicles, robots, and (supposedly) data centres in space.

The eventual goal is to manufacture enough chips to provide 1 terawatt of power per year. Musk argues that current semiconductor makers aren't producing chips fast enough for his AI and robotics businesses, so he's building his own supply.

I'm a bit skeptical of the $119 billion figure though. That's the kind of number Musk likes to throw around, and Grimes County is reportedly only one of several locations under consideration - so the filing could be more of a negotiating tactic rather than a firm commitment.

That said, the underlying logic does make sense. xAI is burning around $1 billion a month just to keep up with OpenAI and Google, and chip supply is the most expensive bottleneck in AI today.

I've no doubt that this new "Terafab" factory will be built, as it's incredibly important for all of his business interests, but whether it's actually built at this scale is another question.

3. AI actors are banned from the Oscars

The Oscars have effectively banned AI from acting and writing roles. To get a nomination, the Academy now requires that performances are "demonstrably performed by humans" and that scripts are "human-authored".

It's a pretty sensible call, given how quickly AI is creeping into Hollywood. It was recently announced that Val Kilmer, who died last year, is being recreated with AI for a lead role in an upcoming film. And one comedian has gone even further, creating an entirely AI-generated actor that she wants to "become a global superstar".

The Academy has stopped short of a wider ban though - AI used for editing, effects, or sound design won't affect a film's chances of nomination. They've reserved the right to ask how AI was used, but there's no clear process for actually verifying this.

I think it's the right decision, even if enforcement will be tricky. Acting and writing are deeply creative roles, and as AI-generated content floods every other medium, audiences are likely to put a real premium on art they know was made by humans. It's important these roles stay valued, rather than always bowing to whatever the technology can offer.

I suppose that rules my own AI short film out of the running then... But if you fancy making one yourself, I've put together a full guide that you can read here.

4. AI models could soon need White House approval before release

The White House is considering vetting new AI models before companies can release them to the public - a major reversal from its previously hands-off approach.

The plan would start with an executive order to create an AI working group of tech executives and government officials, who would design the review process. One option on the table is to follow in the UK's footsteps, where government bodies check that new models meet certain safety standards before they're released.

The shift was reportedly triggered by Anthropic's Mythos model, which the company said was so powerful at finding cyber vulnerabilities that releasing it could cause a security "reckoning". Anthropic has decided against releasing the model for public use.

The White House is now worried that it will be blamed if a devastating AI-enabled cyberattack happens. Public opinion has shifted in the last year, with a Pew Research poll finding that around half of both Republicans and Democrats are now more concerned than excited about AI.

It'll be interesting to see whether the administration actually goes through with this. Just last year, the US Vice President warned that "excessive regulation of the AI sector could kill a transformative industry" - so there's clear disagreement internally about how heavy-handed any review process should be.

5. Smartglasses are secretly being used to film and humiliate people online

Smartglasses are now being used to secretly film people in public, with the videos posted online to embarrass them - and racking up tens of thousands of views in the process.

One woman was approached in a London shopping centre, by a man who was wearing smartglasses. She didn't realise that she was being filmed and only found out after a friend sent her the video, which had already been viewed 40,000 times.

When she asked for the video to be taken down, the man then tried to charge her for the removal. And while the Metropolitan Police did open an investigation, they decided against progressing the case.

Overall, this is a real problem with smart glasses that's only going to get worse. While most smartglasses have small indicators that flash when the camera is recording, people have found ways to cover or disable them entirely.

And with social media making this kind of content lucrative (some videos rack up millions of views and earn creators thousands of dollars), the incentives are firmly stacked the wrong way.

Until these platforms and regulators decide to take action, we should expect a lot more of these stories as smartglasses go mainstream - leaving us with significantly less privacy in an always-on world.

Software engineers are in higher demand than ever, thanks to AI

For years, we have heard about how AI is replacing software engineers. Companies have repeatedly said that they're laying off tens of thousands of workers, so that they can be replaced by artificial intelligence.

But the data from Indeed shows otherwise, as we're now seeing a rapid rise in demand for software engineers. There are several reasons for this, but the number one is the AI gold rush itself.

Companies are now scrambling to integrate the technology into their own products and stay ahead of their competitors. That's created a wave of new features, integrations, and AI-powered tools - which all need engineers to actually build them.

As a result, the very technology that was supposed to replace workers is now generating more work for them than ever. This comes down to the Jevons paradox, which is when something becomes cheaper or easier to do, demand for it tends to increase rather than fall.

This matches what I've been seeing in my own work. AI tools have made me significantly faster at writing code, but I'm not building less - I'm building more, attempting things I wouldn't have tried before, and shipping features that would have taken weeks to develop otherwise.

The bigger question is whether this dynamic holds up over the next few years, or whether the productivity gains from AI will eventually outpace demand for new software entirely. But for now, software developers are in more demand than ever.

200+ use cases for AI, that anyone can use

Just a few days ago, I released a library of 200+ practical ways to use AI in your everyday life - from planning your next holiday, to making sense of medical results, to writing the awkward emails you've been putting off for weeks.

We constantly hear that AI is changing how people live and work, but the how part is harder to find. So I've pulled the examples together myself.

There are over 200 use cases, with each one solving a real problem that you'll recognise:

The library is organised into 24 categories - Travel, Health, Home, Finance, Marketing, Strategy, and 18 more - so that you can skip straight to what's relevant for you right now.

Every use case includes a prompt that you can try yourself and a real example response. And it works with whatever AI tool you already use, whether that's ChatGPT, Claude, or Google Gemini.

It's completely free, and you can start using it immediately.

🚀 OpenAI partners with rivals to upgrade AI supercomputer networking

💰 Ramp could hit a $40 billion valuation, 6 months after reaching $32 billion

⚓ Peter Thiel backs a startup building floating data centres

📈 Kalshi doubles its valuation to $22 billion in just 5 months

💪 Google launches an AI health coach for $9.99 a month

🇨🇳 Moonshot AI raises $2 billion at a $20 billion valuation, as Chinese open source AI booms

🤝 Anthropic signs a major compute deal with SpaceX, leads to higher usage limits for Claude

📊 AMD's revenue jumps 38% as Q1 data centre sales hit $5.8 billion

📱 Samsung hits a $1 trillion valuation thanks to the AI boom

🏠 Google Home can now handle more complex requests with Gemini

🎨 Image models are now driving app growth, rather than chatbot upgrades

🏢 Sierra raises $950 million as the enterprise AI race heats up

If your team has ever built an AI chatbot, they will have known the pain of getting everything setup and working for different screen sizes. This means that companies can spend weeks on perfecting the user interface from scratch, rather than actually testing their AI tool with customers.

Copilotkit is a company that has just raised $27 million in Series A funding and has been building a library of re-usable UI components for the last few years. They've started to gain traction recently and have launched their own protocol, which standardises how AI agents talk to user interfaces and each other.

This is tapping into an emerging trend called Generative UI, which is when we give AI agents the ability to create interfaces within the chat window - such as buttons, forms, or interactive dashboards.

It's a much more interactive experience for users, compared to constantly reading walls of text (which most people skip over anyway), and makes it easier for us to adopt these AI tools.

Of course, it's still early days and many companies are still trying to figure out what this should look like - but it's something that I've been flagging in this newsletter for over a year. Generative UI is the next big thing that companies should be focused on.

It's great to see CopilotKit raise this kind of money. I've been using Copilotkit's UI library for well over a year, especially when it comes to my professional work, and they've done a pretty good job here. You can easily get a working prototype setup in minutes and add your own company branding.

Most of the top agentic AI frameworks are supported, including LangChain's Deep Agents - which is my go-to for creating custom AI agents.

If you want to find out more about the company and what they're doing, I've included a link below.

This Week’s Art

We’ve covered quite a bit this week, including:

If you found something interesting in this week’s edition, please feel free to share this newsletter with your colleagues.

Or if you’re interested in chatting with me about the above, simply reply to this email and I’ll get back to you.

Have a good week!

Liam

Share with Others

If you found something interesting in this week’s edition, feel free to share this newsletter with your colleagues.

Liam McCormick is a Senior AI Engineer and works within Kainos' Innovation team. He identifies business value in emerging technologies, implements them, and then shares these insights with others.