← Tech Brew Ride Home
Tech Brew Ride Home
New Claude, New Realities
00:24:10
Anthropic dropped Opus 4.8 with dynamic workflows for Claude Code and raised $65B at a $965B valuation, overtaking OpenAI. Blue Origin's New Glenn exploded during testing, Amazon killed its AI usage leaderboard, and an AI startup offers free home cleaning for training data.
Loading summary
Transcript5 lines
- [00:00]
A
Study and play come together on a Windows 11 PC and for a limited time, college students get the best of both worlds. Get the unreal college deal Everything you need to study and play with select Windows 11 PCs. Eligible students get a year of Microsoft 365 Premium and a year of Xbox Game Pass ultimate with a custom color Xbox wireless controller. Learn more@windows.com studentoffer while supplies last ends June 30th terms at aka mscollegepc.
- [00:35]
B
Welcome to the Tech Brew Ride home for Friday, May 29, 2026. I'm Brian McCullough. Today, Anthropic dropped Opus 4.8 and raised $65 billion at a greater than OpenAI valuation. Blue Origin's new Glenn Rocket exploded during testing. Amazon killed its AI usage leaderboard. An AI startup is offering free home cleaning for training, data and of course, the weekend Long Read Suggestions here's what you missed today in the world of tech well, Claude Opus 4.8 is here, and I'll tell you why that's interesting in just a sec. But first, quoting the Verge, Anthropic is releasing Claude Opus 4.8 on Thursday, and the company is touting the model's honesty. According to Anthropic, it trains all its models to be honest, for instance, to avoid making claims that they can't support. But it notes that quote, a general problem with AI models is that they sometimes jump to conclusions, confidently presenting their work as making progress despite thinking in evidence. The AI lab claims that early testers have found that Opus 4.8 is more likely to flag uncertainties about its work and less likely to make unsupported claims. In the company's evaluations, Opus 4.8 is around 4x less likely than its predecessor to allow flaws in code it's written to pass unremarked. In addition to the honesty, improvements with Opus 4.8, users can direct the amount of effort Claude puts into a task. Higher effort responses will use more tokens, giving users the option of lower effort responses if they don't want to burn through their rate limits as quickly. So it's interesting that this is out already because this is the quickest that Anthropic has released a model upgrade ever. They sort of had to though, right? For reasons we've discussed. Quoting Bindure on Twitter, Opus 4.8 still trails behind GPT 5.5 and is a very incremental release. Opus 4.8 barely inches past 4.7 on benchmarks, but lags behind GPT 5.5 considerably. Anthropic may be be stalling a bit given its last two releases. OpenAI has a huge opening with GPT 5.6 coming soon. End quote yeah, Anthropic must be feeling the pressure to keep pace, which is why they probably pushed this out as an attempt to leapfrog. And by this I mean this Quoting Testing catalog Anthropic has announced the introduction of dynamic workflows within Claude Code, a feature designed to handle complex engineering tasks at scale. This new capability allows users to orchestrate large multi step coding projects through parallelized agents that operate on subtasks simultaneously. For example, the porting of BUN from Zig to rust, involving around 750,000 lines and rigorous test suite validation was accomplished by leveraging these workflows. The system automatically plans and distributes work, checks results for accuracy, and iterates until consensus is reached, ensuring coordination even for jobs that span several days. Users are prompted for confirmation before a workflow executes and and organization admins can manage access and settings. Anthropic's CLAUDE code platform, known for its AI powered coding assistance, is now targeting power users including engineering teams and organizations managing large scale projects. Dynamic workflows are immediately available to those on Max and team plans or via the API, with enterprise customers able to opt in by enabling the feature through admin controls. This launch positions Anthropic competitively against other AI coding assistants by scaling to more demanding work and offering features such as workflow recovery and granular administrative oversight. Early user feedback highlights the tool's ability to expedite previously time intensive engineering processes. End quote so essentially get ready to pay Claude a lot more money because essentially you can do this. Your model is set to opus 4.8. Then you set your reasoning effort to ultracode to enable dynamic workflows like I did this morning, and CLAUDE will automatically detect your complex tasks, then write an orchestration script and then spawn an agent swarm. Hundreds of parallel agents that go out, do the work and check each other's work. Quoting Nick Dobos on Twitter Cloud Code's new Dynamic Workflows update is absurd. Make sure you understand what it's doing here. This isn't simply a long running mode like Goal or a fancy subagent verifier process. This is Claude Vibe coding an entire brand new sub agent fleet harness on demand RLM on agent harness. That's what dynamic workflow means. This is basically a new scaling law dimension. End quote Anthropic also said it expects Mythos class models to be available to all customers in the coming weeks following the development of stronger safeguards. So maybe this all really is just a stopgap for now. Oh, and there's also this quoting the Times On Thursday, Anthropic punctuated its ascent by officially passing OpenAI as the world's highest flying AI startup. Anthropic said it had raised $65 billion in financing that values it at $900 billion before the inclusion of the new capital, a deal that puts it ahead of OpenAI's last valuation of 730 billion. The company also unveiled a new flagship AI model. Cloud Opus 4.8 is significantly better than its predecessor at generating computer code. The new investment, which was led by investors including Green Oaks Capital, Sequoia Capital, Altimeter Capital and Dragoneer Investment Group, boosted Anthropic's value to nearly two and a half times its previous valuation of $380 billion about three months ago. In its latest funding, Anthropic brought in new strategic investors, including Samsung, Micron and Skhenix, companies that build storage, memory and logic chips that are crucial to the development of AI. In a blog, Anthropic said these strategic partners would help boost the company's computing power as demand for CLAUDE code grew. Anthropic has now raised more than $130 billion since it was established, according to Pitchbook, which tracks startups. End quote. Bad news over in Bezos ville Quoting the FT Blue Origin's New Glenn rocket exploded during testing on Thursday, days before it planned to launch Internet satellites for Amazon and a further setback for J. Bezos Space Co. The company said it had experienced an anomaly during today's hot fire, a pre flight procedure in which the rocket was loaded with fuel and its engines were fired to test systems. Blue Origin said all personnel were accounted for and it was investigating the incident, multiple employees told the FT an initial assessment of the site showed severe damage to Blue Origin's only launch pad and other equipment, including a lightning tower and transporter erector used to lift rockets into position ahead of launch. The failure was expected to severely impact the company's launch plans this year, the people added. Very rough day, bezos posted on X. But we'll rebuild whatever needs rebuilding and get back to flying. It's worth it. End quote New Glenn was scheduled to ferry 48 satellites for Amazon's Leo Internet constellation on Monday. Video footage from the launch pad in Florida showed the rocket exploding into a fiery mass on Thursday night. Blue Origin has been seeking to build up its launch cadence and plan to carry out 12 flights using new Glenn Year. It is aiming to compete with Elon Musk's SpaceX, which is on course for a record initial public offering next month. Most unfortunate, Elon Musk posted on X after the new Glenn explosion. Rockets are hard, end quote. Sources say that Amazon has shut down that internal leaderboard that tracked employees usage of AI tools after Amazon workers tried to boost their scores on the leaderboard with, you know, needless tasks Quoting the FT employees at Amazon were told this week that its Kiro Rank service, which scored users of Amazon's Kiro developer platform based on their AI activity, had been taken offline, according to two people familiar with the matter. The decision came after the tool led some workers to assign AI agents autonomous bots that can take actions on behalf of users to carry out needless tasks in an apparent attempt to to climb the rankings. Dave Treadwell and Amazon senior vice president told staff earlier this week that the leaderboard had been built with good intentions, according to people familiar with his remarks. But he added that the result had been additional costs for Amazon due to employees token maxing or inflating their consumption of AI tokens, units of data processed by models. Please don't use AI just for the sake of using AI, he told staff. Amazon confirmed in a statement that the beta dashboard was not a formal or approved tool and has since been deprecated. The move highlights how tech groups efforts to encourage the use of AI can lead workers to try to game performance measures with pointless activity increasing infrastructure expenses. Meta employees have similarly sought to boost their position on internal tables by driving up token consumption. Amazon's decision comes after the FT reported how staff were using Kiro and meshclaw, an in house version of the popular openclaw tool that allows users to run agents on their own hardware. Some employees said colleagues were using the software to generate additional AI activity to drive up token consumption and demonstrate adoption of the technology. The behavior comes amid growing pressure on staff to adopt the technology after Amazon introduced targets for more than 80% of developers to use AI each week. End Quote.
- [10:32]
C
Ready to soundtrack your summer with Red Bull Summer All Day Play? You choose a playlist that fits your summer vibe the best. Are you a festival fanatic, a deep end dj, a road dog or a trail mixer? Just add a song to your chosen playlist and put your summer on track. Red Bull Summer All Day Play Red Bull gives you wings. Visit red bull.com brightsummerahead to learn more. See you this summer. Starting a business can seem like a daunting task unless you have a partner like Shopify. They have the tools you need to start and grow your business. From designing a website to marketing to selling and beyond, Shopify can help with everything you need. There's a reason millions of companies like Mattel, Heinz and Allbirds continue to trust and use them. With Shopify on your side, turn your big business idea into Sign up for your $1per month trial@shopify.com SpecialOffer the Wired
- [11:28]
D
newsroom is known for award winning reporting on how technology shapes our world. On WIRED's Uncanny Valley, we take that curiosity even further. Each week, journalists from Wired break down the biggest stories in tech while speaking directly with the people building challenging and reshaping the future Is the AI boom sustainable? How do you protect your privacy in an age of constant surveillance? Uncanny Valley tackles the questions driving today's tech debates and lighting up your group chats. Listen to new episodes every Thursday, wherever you get your podcasts.
- [12:00]
B
So would you let them do this? I don't know that I would, Quoting the Verge AI training startup Shift wants to clean your house for free. The catch, because despite what its website says, there's always a catch, is that it will record cleaners as they scrub, vacuum, dust, tidy and wash, and use that footage to train robots. Shift announced the unusual offer on social media on Thursday, explaining that the value of the training data generated from the cleanings is more than enough to fund the service. As its website puts it, you get a spotless apartment, we get training data. Everyone wins. A promotional video shows a cleaner in a crisp white uniform, an awkward looking hat. More on that later. Washing windows, mopping and vacuuming floors, scrubbing dishes and wiping down counters. According to Schiff's co CEO and co founder Burkan Kilik, this magic hat is what records the work peak fashion it is not, but it does contain a camera that captures footage from the cleaner's point of view. Footage from inside your home is, of course, what you're paying for the cleaning service with. On its website, Shift says customers privacy is fully protected with sensitive details like names, faces or personal information from screens and ID cards blurred and anonymized before being used for AI training. Shift says its cleaners are also vetted by its partners, though stresses they are not Shift employees. Every home cleaned today lays the groundwork for a home that cleans itself tomorrow, the company says in the video. As it happens, the dirtier the better. An FAQ on the company's website says more challenging cleaning environments can be especially useful. There are limits, however, and cleaners may decline any specific tasks they are not comfortable performing. The service is initially only available in New York, but Kilik says it will be available very soon in San Francisco, London, Zurich and Munich. The free cleanings are only available for a limited time, but the model fits within a growing market for recording of human tasks that can be used to train AI systems and robots. Shift says if it already pays tens of thousands of people across 15 countries to record their activities through its app, cleaning may only be the start. Schiff's video says it eventually plans to move into other areas like plumbing, cooking and building. End Quote. Time for the weekend Long Read suggestions and first up, Simon Willison sums up what is going on in AI right now. He says, essentially, anthropic and OpenAI seem to have finally found product market fit with coding agents, which are quickly becoming the daily drivers for very well paid professionals. Quote I currently subscribe to the $100 a month max plan from Anthropic and the $100 a month pro plan from OpenAI. If you are a heavy user of coding agents, these plans are a fantastic deal. I just ran the CC Usage tool on my laptop to get an estimate of how much I would have spent if I were to pay for API tokens in the past 30 days and got $1,199.79 for anthrop code, $980.37 for OpenAI codecs. That's $2,180.16 worth of tokens for $200. Not bad at all. I'm a moderately heavy user of these tools, but I'm certainly not running agents every hour of the day and night. I had assumed that companies making extensive use of agents were getting similar discounts. It turns out I could not have been more wrong about that. I haven't been able to track down the exact date, but at some point in the the last six months Anthropic switched their enterprise plan originally. Claude seats include enough usage for a typical workday back in August 2025 to $20 per seat per month plus API pricing for usage. This story about the change from the Information is dated April 14, 2026, but cites an Anthropic spokesperson claiming that the pricing change occurred in November 2025. Existing customers are finding out about the change as they renew their contracts. OpenAI made a similar pricing change in April, touching on and adding API pricing on top of a monthly fee. Why these sudden aggressive moves on pricing? Both Anthropic and OpenAI are planning to IPO, but I suspect there's a more important factor here. I think they finally found a product market fit with the coding and general purpose agent products embodied by Claude Code and Cowork and codex. Tools like ChatGPT are wildly popular, but that wild popularity has been difficult to turn into revenue. In February, OpenAI boasted more than 900 million weekly active users for ChatGPT, but 50 million, or 5.6% of that were paying consumer subscribers. Charging $10 to $20 per month per user is an okay business, but you need 1 to 2 billion subscribers sticking around for four years to cover $1 trillion in infrastructure companies. Spending $200 plus per month per user will get you there a whole lot faster. And as noted above, as a power user I'm at $1000 per month in API costs per vendor already. Coding agents really did change everything. These are tool vastly more tokens, but are also quickly becoming daily drivers for the work carried out by extremely well compensated professionals. Right now that's still mostly software engineers, but a coding agent is a tool that can automate anything you can do by typing commands into a computer, so they are clearly applicable to a much wider set of skilled knowledge workers. As I've discussed on this site at length, the models released in November 2025 elevated agents to being genuinely useful. We've had six months to get used to that idea. Now it's no wonder companies are beginning to spend real money on this technology. End quote. And I'm connecting this second long read to that one. The Financial Times says that Kirkland and Ellis, the world's highest grossing law firm, is setting aside $500 million to build its own AI platform rather than rely on tools available from the larger platforms but also available to its rivals. Quoting the ft the idea is that we're going to take the collective intelligence of our institution and be able to deploy that throughout our firm, a representative said. While the use of widely available AI tools was raising the floor for everyone in the legal industry, the spokesperson said Kirkland had to do more than use those because we don't get hired for the floor. Outside companies were building the Kirkland technology alongside Kirkland's own engineers and data scientists, the representative said. But they would not be able to sell it to others, and the law firm would own or have the right to own all of it, he said. 180 tech professionals were working on it. In total, the AI platform was being designed using information from a group of 250 Kirkland lawyers, including 100 partners, about how they did their jobs, he said. Kirkland will pay for the investment out of its revenues, eating into the profits available to be shared among equity partners, at least in the short term. The firm reported a record $10.6 billion in revenue last year. The AI platform will be used for the firm's mandates in their entirety, using the knowledge that partners have into it, instead of lawyers calling on individual programs for specific tasks, he said. The name of the platform and the technology companies involved are set to be disclosed in the coming weeks, end quote. So, bunch of random thoughts on that in random order, but they're all connected. Yeah, individuals like me have realized they can now create any product that they need for their daily life in software. But businesses can realize that too. Even small businesses like Facebook doesn't pay for Slack. Facebook has its own internal version of Slack, but they're Facebook. They're big enough and technical enough to be able to build their own internal tools when they need them. What if your say, four person accounting firm could code up an internal version of Slack just for you, just over a weekend? Then you wouldn't have to pay your Slack subscription. I mean, if you can do it yourself, why pay any subscriptions for a SaaS tool if you could cheaply and easily clone the ones you use for yourself and would have the added benefit of being 100% tailored to just your business's needs? People have been fearing the sasspocalypse because they feared it was about the big AI model companies just routinely releasing features that would eat the lunch of the existing SAS incumbents. But what if the real SaaS apocalypse is that agents and vibe coding could empower any enterprise of any size to just build it all themselves? So number one, Arsenal is in the Champions League final tomorrow afternoon. Watch it. We are about to become Champions of Europe for the first time. And number two, I've got a bonus episode for you this weekend. It's a portfolio profile episode on a company the Right Home Fund has invested in that is shocker, not AI. It's a really interesting company in the sense that it has already just in the go to market phase, saved multiple people's lives. Saved actual human lives. Sound interesting? Listen to the episode. Talk to you on Monday.