You’re receiving this email because you subscribed to the newsletter. If you’d like to unsubscribe, you can manage your preferences here with one click (or use the button at the bottom of the email).

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!

1. Amazon reveals AI agents that can code autonomously for days

We start this week with Amazon, which has unveiled three "frontier agents" that can autonomously write code and deploy new software. Kiro is one of these AI agents and Amazon claims that it can work independently “for days”.

All three agents work together and handle specific tasks - with Kiro focused on coding, the security agent then reviews the code and spots vulnerabilities, and the DevOps agent finally deploys the software.

Interestingly, Kiro can scan your code base and learn the coding standards that your company follows. This should lead to better results and more closely match the work that developers produce.

Essentially, this is Amazon’s answer to Claude Code - a tool that developers have quickly adopted and now use to generate their code.

But Amazon is focused on enterprise customers instead, who need security guarantees over what the agent can and can’t do - along with an easy way to deploy and monitor the agent.

I’ve been building AI agents for over two years now and the technology has come a long way, but there’s a lot further to go. This is one step towards accelerating the software development process, but developers are still needed to check and verify the results.

Practically speaking, it’s impossible for us to give these AI agents a complete picture of our project and all the requirements. We cannot write down every single requirement, as it takes too long, some can be forgotten, and some are implicit - which is a problem for AI systems.

As a result, AI agents can struggle to plan effectively and their performance degrades when you try to use them for longer running tasks. That won’t suddenly change with Kiro, so companies should use it for shorter tasks instead.

2. Waymo approved to expand across Bay Area and Southern California

In the last few months, Waymo has announced a string of new locations and continues to expand its robotaxi business. Regulators have now approved plans for it to operate robotaxis across the Bay Area and Southern California - allowing it to test across a significantly larger area.

The company, which is owned by Google, already runs services in San Francisco, Silicon Valley, LA, Phoenix, Austin, and Atlanta. By expanding their fleet and testing them over wider areas, Waymo is able to collect more data and improve its AI algorithm.

For over a decade, the company has been running their vehicles in downtown San Francisco and gradually ramped up the service. But there has been a substantial shift in Waymo’s strategy this year, as their executives have become more confident in the technology’s ability to scale and announced plans to expand rapidly.

The company has always been more cautious than its rivals, which allowed them to build more trust in the technology and get buy-in from regulators - who are understandably risk-averse with self-driving cars.

That cautious approach has worked well for Waymo. Their nearest competitor was Cruise, until the company faced a string of negative headlines and eventually caused an accident - with one person dragged under the car.

California then revoked Cruise’s permits and the company was shut down a year later. But Waymo was more careful in how it expanded and much open with regulators when things went wrong, which is why they’ve just been given this approval to expand across California.

3. Startup has found a huge lithium deposit, thanks to satellite images and AI

As an industry, we often focus on the huge tech companies and their latest research - or the worrying ways that technology can be used to cause harm. But there are plenty of innovative ways that the technology is being used, beyond what the large companies are focused on.

One of those is Fleet Space, which is an Australian startup. The company has combined satellite imagery with its own AI platform, leading to a new mineral deposit being discovered in Canada.

To do this, the company operates satellites and uses their electromagnetic sensors to outline what’s beneath the Earth’s surface. This data is then sent to their own AI platform, which is able to recommend new drilling locations in 2 days.

This has allowed the team to accelerate the discovery process, in an industry where only 0.3% of deposits are commercially viable and companies had to spend weeks doing analysis - that has now been cut down to just 48 hours.

As a result, the Cisco lithium project in Quebec has already been expanded and it’s believed to contain 329 million metric tons of lithium oxide. It comes at an important time for the wider tech industry, as electric vehicles and new devices are leading to a surge in demand for lithium.

4. OpenAI and Anthropic are preparing for IPO

Both AI startups are in early discussions about an IPO launch, which could happen as early as next year. Anthropic has brought on Wilson Sonsini Goodrich & Rosati - the law firm that has previously organised IPOs for Google, LinkedIn, and Lyft.

At the same time, Anthropic is pursuing private funding that could push its valuation beyond $350 billion, with Microsoft and Nvidia reportedly committing a combined $15 billion.

This follows similar moves from OpenAI, which could lead to both companies racing to market and have ramifications for the wider industry. It’ll be an interesting gauge for investor confidence in the sector, or could raise concerns about an AI bubble and dampen enthusiasm.

An Anthropic spokesperson was careful to downplay the speculation, noting that companies at their scale typically operate as if they're already public, and that no decisions have been made.

Still, the signs are there.

The company has recently hired Krishna Rao, a former Airbnb executive who led their 2020 IPO. He has been managing the company’s $50 billion infrastructure build and is leading efforts to triple its international headcount.

Recent valuations have already placed Anthropic around $350 billion, while OpenAI’s funding rounds have placed it at $500 billion. I think it’s only a matter of time until they launch an IPO.

5. Black Forest Labs raises $300 million to develop new image models

The German startup has quickly established itself in the image generation space, with its AI models frequently being used by developers and top companies - including Adobe, ElevenLabs, VSCO, and Elon Musk’s X.

I flagged this company last year, which you can read in more detail here, and they’ve grown dramatically since then.

The startup has just closed a $300 million Series B round, pushing its valuation to $3.25 billion just one year after it launched. With this fresh funding, Salesforce, a16z, NVIDIA, Canva, and Figma have joined as investors.

The $300 million will fund further R&D, which likely means more model improvements and potentially new capabilities beyond static image generation.

Following their recent release of Flux 2, it’s clear that the technology is heading towards 4K quality and more consistent image generation - which is a common issue with today’s technology.

100 ways to advance your career with AI (prompt pack)

In case you missed it, I created a prompt pack that can help you advance your career - from landing your next role, to negotiating a better salary, to positioning yourself for promotion.

It includes over 100 prompts covering everything from career planning to skill development, with plenty of examples to get you started:

You don't need to learn anything new. Just click a prompt and it opens directly in ChatGPT - or if you prefer Claude or Gemini, the prompt is automatically copied to your clipboard and ready for you to paste it in.

The prompts are organised into stages - including Plan, Skills, Apply, Interviews, Negotiate, Branding, Networking, Promotion, and Transition - so you can jump straight to what matters for you right now.

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

Let me know what you think - I'd love to hear which prompts you find the most useful. Or if you have any questions, simply reply to this email and I’ll get back to you.

LangChain releases low-code tool for building AI agents

LangChain has opened up its Agent Builder to everyone, allowing people to create AI agents in minutes - even if you have no coding experience.

The process is quite straightforward and similar to Microsoft’s Copilot Studio, which is another low-code tool that allows you to create AI agents.

However, LangChain’s tool is simpler to use and can be given access to your Gmail account - which isn’t something that Microsoft will offer.

To get started, you simply interact with their AI chatbot and provide it with access to your OpenAI/Anthropic API keys. It then builds the agent for you and asks a series of questions.

With LangChain’s tool, you can connect the agent with your own data and external APIs, choose between different AI models, and call the agent programmatically.

There's also a new workspace that allows teams to view and customise each other's agents, so that someone's great idea isn’t hidden away and can be used throughout your company.

Sales teams are already building agents that can research prospects before their morning calls. Product teams have also used it to create bots, which are able to turn Slack messages into Linear tickets - including details like the scope and priority.

I've got direct experience building AI agents for sales teams, and getting good results can be tricky. These agents are powered by generative AI models, which means they will occasionally slip up or become confused.

Sometimes we can get excited by this technology and forget about the limitations, which can lead to some embarrassing mistakes. If you're relying on an agent's output for a sales call, make sure that you double-check the information beforehand.

That aside, it’s fantastic to see that LangChain has released this tool and made it widely available for the public. If you haven’t used it already, I highly recommend that you give it a try - it’s free to use and a good way to build your own AI agents.

🎬 Runway releases a new video generator that competes with OpenAI and Google

🤖 Google announces a central platform for companies to develop AI agents

🚕 Uber launches robotaxis in Dallas and the UAE, while Waymo has started testing in Philadelphia

📉 ChatGPT's user growth is starting to slow down

🚀 Anthropic releases Claude Opus 4.5 as its flagship model

📰 Meta signs AI licensing deals with CNN, Fox News, and USA Today

👔 Apple AI chief steps down, several other executives depart

🚗 Nvidia's latest tools will support autonomous driving research

⚖️ New York Times sues Perplexity for copyright infringement

🎭 Netflix will acquire Warner Bros. for $82.7 billion

This startup from Singapore has secured $110 million to support its merger with dMY, exceeding its original target by more than 120%.

The funding round attracted heavyweight backers including IonQ (one of the world's largest quantum computing firms), a Fortune 50 tech company, and several institutional investors.

Once the merger is completed in the next few months, Horizon will have roughly $137 million cash on hand.

You might have read about them in the last week, as they became the first private company to run a quantum computer for commercial use. The company was founded in 2018 and it’s building the tools and infrastructure that power quantum applications.

Software engineers can use Horizon’s IDE to create their own quantum applications, without having to worry about the hardware requirements.

While much of the quantum world is quite research-focused and has a long way to go, this is a necessary step to get it into the hands of developers and allow companies to solve real-world problems - whether they’re in pharmaceuticals, finance, or other industries.

For Singapore, this cements its ambition to become a regional hub for quantum research - with the government planning to spend S$300 million on its National Quantum Strategy.

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.