1/ As consumer agents start to take off with Muse, Instinct, and others to come We wanted to build a tool that would help researchers understand what’s actually happening under the hood We were especially motivated by reports of Instinct losing users’ money Here’s a demo of our Interceptor tool with Muse:

Just read this piece from my alma mater about a post-training startup that a lot of students have been working for to make side money. I found it interesting. In some sense there's a dystopian element to it: students training their replacements. There's also an eerie similarity to gig work like Uber and DoorDash. This time it's for true knowledge work, which reinforces a lot of the fears people have about the future of whitecollar work, especially entry-level work. On the other hand, is it crazy to believe data annotation is a category that gets a lot bigger from here? That a big portion of our economy will be focused on training LLMs? As long as there's interesting stuff to measure and things humans know that AI doesn't, there should be demand for this kind of data. This could be a growth category. Plus, I'm not convinced that the "AI takes all jobs" narrative is correct, and if anything, gaining fluency in working with LLMs could be an advantage for going into the workplace. For a student, a lot of these roles pay pretty good money. Plus, you can do it from your bedroom. Interesting category of companies to keep track of. h/t @zile_cao for sharing this article with me.

We're backing @tarecredit because the credit system is old, expensive, and ready for an upgrade. Credit is heavily intermediated with slow and opaque manual processes and the plumbing alone costs borrowers and investors north of $100B a year. It's finally become possible to rebuild the entire value chain around a single source of truth, using stablecoins, smart contracts and AI. Led by co-founders @humancapitol @lucasvo and @keerthimoudgal, Tare is the perfect team for the job. Kevin helped build one of the first onchain private credit funds at Blocktower Credit after an extensive career in credit before, Lucas co-founded Centrifuge and is a pioneer in RWAs, and Keerthi was Head of Product at JP Morgan's institutional blockchain platform Kinexys. We're thrilled to announce that we led their seed round alongside a great group of funds and angels including leading private credit firms like Apollo and Janus Henderson and founders like @StaniKulechov of Aave. @jonah_b and I wrote more about our thesis below.

Nasdaq Ventures is investing $100M in Payward as we advance tokenization together with the @Nasdaq Equity Token (NET) framework. Equities on-chain only works if it's built to institutional standard from day one. We're building it with the company that's run public market infrastructure for decades. → payward.com/press-release/pa…

Nasdaq Ventures is investing $100M in Payward as we advance tokenization together with the @Nasdaq Equity Token (NET) framework. Equities on-chain only works if it's built to institutional standard from day one. We're building it with the company that's run public market infrastructure for decades. → payward.com/press-release/pa…

Today, we are at a crossroads. As I write this, AI is saving lives by helping doctors identify diseases faster, and making education more accessible to people everywhere. It’s also being used to inflict harm on a global scale. TRM’s AI-in-Crime Adoption Index found that, in 2026, criminal adoption of AI rose 40% year over year. Today’s threat actors are using AI to execute scams, cyber attacks, and other forms of digital crime at unprecedented speed and scale. In the last 90 days, we’ve also seen a growing number of incidents where AI agents themselves have gone rogue. Business titans and religious leaders are sounding the same alarm: to realize AI’s promise, we need to protect people from its harms. The first line of defense is prevention. Model companies like Anthropic and OpenAI are building stronger guardrails, and governments are exploring rules for how AI can be safely used. We call this the prevention layer. At TRM, we’ve spent nearly a decade fighting crime, from fraud and cybercrime to human and wildlife trafficking. That experience has taught us one thing, if nothing else: the prevention layer is essential, but never enough. Criminals will break through. When they do, it’s critical that crimefighters can identify the harm done, find the people and machines behind it, and take action: support the victim, freeze the funds, take down the infrastructure, arrest the offender, disable the agent, dismantle the network. The world needs a new layer of protection. We call it the disruption layer—and it's what we're building, right now. The operating system for crimefighters In 2018, we founded TRM to help crimefighters map how threat actors were using cryptocurrency. We analyzed petabytes of data, mapping over a 100 types of crime. Government agencies and financial institutions worldwide now use TRM to investigate complex criminal networks and combat financial crime. The more cases we supported, however, the more we realized: if we were going to equip crimefighters to disrupt AI-scale crime, crypto would be 𝘱𝘢𝘳𝘵 of the story, but not the 𝘸𝘩𝘰𝘭𝘦 story. To build a successful case, investigators must source intelligence not just from the blockchain, but also corporate records, cyber threat data, the dark web, and dozens of other domains. Today’s crimefighters need to use natural language to search across siloed domains, connect the dots and unearth buried insights, automate the rote work, and coordinate with partners across public and private sectors—all while preserving auditability and human judgment. In short, crimefighters need an operating system for the AI era. So we built it. For years, we’ve been developing this OS alongside the investigators who use it. Now, we’ve just announced a new round of funding, valuing TRM at $2 billion, to propel our work—and this fall, we’ll unveil the OS publicly, for the first time. If your organization needs to fight AI-scale crime, please reach out. We want to work with you.

Today, we are at a crossroads. As I write this, AI is saving lives by helping doctors identify diseases faster, and making education more accessible to people everywhere. It’s also being used to inflict harm on a global scale. TRM’s AI-in-Crime Adoption Index found that, in 2026, criminal adoption of AI rose 40% year over year. Today’s threat actors are using AI to execute scams, cyber attacks, and other forms of digital crime at unprecedented speed and scale. In the last 90 days, we’ve also seen a growing number of incidents where AI agents themselves have gone rogue. Business titans and religious leaders are sounding the same alarm: to realize AI’s promise, we need to protect people from its harms. The first line of defense is prevention. Model companies like Anthropic and OpenAI are building stronger guardrails, and governments are exploring rules for how AI can be safely used. We call this the prevention layer. At TRM, we’ve spent nearly a decade fighting crime, from fraud and cybercrime to human and wildlife trafficking. That experience has taught us one thing, if nothing else: the prevention layer is essential, but never enough. Criminals will break through. When they do, it’s critical that crimefighters can identify the harm done, find the people and machines behind it, and take action: support the victim, freeze the funds, take down the infrastructure, arrest the offender, disable the agent, dismantle the network. The world needs a new layer of protection. We call it the disruption layer—and it's what we're building, right now. The operating system for crimefighters In 2018, we founded TRM to help crimefighters map how threat actors were using cryptocurrency. We analyzed petabytes of data, mapping over a 100 types of crime. Government agencies and financial institutions worldwide now use TRM to investigate complex criminal networks and combat financial crime. The more cases we supported, however, the more we realized: if we were going to equip crimefighters to disrupt AI-scale crime, crypto would be 𝘱𝘢𝘳𝘵 of the story, but not the 𝘸𝘩𝘰𝘭𝘦 story. To build a successful case, investigators must source intelligence not just from the blockchain, but also corporate records, cyber threat data, the dark web, and dozens of other domains. Today’s crimefighters need to use natural language to search across siloed domains, connect the dots and unearth buried insights, automate the rote work, and coordinate with partners across public and private sectors—all while preserving auditability and human judgment. In short, crimefighters need an operating system for the AI era. So we built it. For years, we’ve been developing this OS alongside the investigators who use it. Now, we’ve just announced a new round of funding, valuing TRM at $2 billion, to propel our work—and this fall, we’ll unveil the OS publicly, for the first time. If your organization needs to fight AI-scale crime, please reach out. We want to work with you.

excited to open source AgentFormation it's a template for spinning up remote, secure, and browser-accessible workspaces for coding agents in AWS CloudFormation. I've been using it daily while walking to work or before heading to bed to kick off a coding task from my phone although we first built this over 8 months ago before claude/codex's remote-control features were released, I still find it great if you prefer TUI's over chat interfaces disclaimer: AgentFormation is still experimental so take a second to review before using it in prod github.com/BlockchainCap/Age…