Nov 10, 2026 · 7 min listen · Last updated November 10, 2026
From storyflo. This is your daily audio brief. It's Theo. November 10th, tech roundup — five stories, here's number one. Let's get into it. First, from AI by Aakash. Spotify's Claude Code Token Trick, Tested: The Real Savings. Spotify's approach to coding with AI is pretty fascinating.
Listen · storyflo · tech
Daily Tech Brief · November 10th
0:00-7:05
Pick your daily storyteller
Subscribe to match with Theo, Jessica, Chloe, Mason, Brock — your voice, every brief.
Audio pre-rendered by Storyflo · cached + delivered from the edge
Spotify's Claude Code Token Trick, Tested: The Real Savings
Spotify's approach to coding with AI is pretty fascinating. Their engineers have been relying on Claude Code so much that many haven't even written code since last December. They’ve streamlined bug fixes to the point where an engineer can call Claude from their phone during their commute and push updates live before they even reach the office. However, this convenience has led to some hefty costs. A project manager at Spotify managed to cut Claude's bill by up to 90% using a clever plugin called Shunt, which prevents Claude from reading large files in full, opting instead for a cheaper model to handle the heavy lifting.
What’s interesting is how Spotify’s internal tool, Honk, has transformed their workflow. It can merge an impressive 1,000 code changes every ten days, a task that used to take three months. By implementing Shunt, they’ve seen significant reductions in token usage, especially on large files, with reductions of 82% to 94%. They even open-sourced the plugin, so others can benefit from their findings.
When I looked into the costs associated with using AI for coding, it was eye-opening. A staggering 94% of the bill comes from reading files, while output costs are surprisingly low. This means that optimizing how models read files can lead to substantial savings. I ran tests on different codebases and found that while using Spotify's method yielded significant savings on one repository, it was less effective on another, highlighting that results can vary based on the specific context.
Overall, Spotify’s innovative use of AI in coding not only showcases how companies can leverage technology for efficiency but also offers a roadmap for others looking to cut costs without sacrificing quality.
California’s proposed ‘billionaire tax’ fails to pass
California voters decided against Proposition 40, which aimed to impose a one-time 5% tax on billionaires. The campaign saw a significant financial imbalance, with tech industry leaders pouring in about ten times more money to defeat the measure than its supporters managed to raise. It's interesting to note that opinions among economists and founders were divided on the potential impact of the tax. Some believed it could push wealthy residents to relocate, while others thought it might only reduce their wealth without driving them away. This outcome highlights the ongoing tension around wealth and taxation in the state, especially in the tech sector.
A New AI Tool From Google DeepMind Proved the Signal Protocol Is Secure
Google DeepMind researchers used agentic AI to mathematically prove, or “formally verify,” a number of the Signal protocol’s key cryptographic guarantees—a landmark development for the widely used cryptography and AI itself. Conducted as a sort of proof-of-concept project in just four weeks, the work is significant in two ways. It serves as further independent validation of the cryptography that protects massive quantities of the world's communications on platforms like the Signal app, Google Messages, and WhatsApp.
Would I Unplug Again? My Experience with Ancient Medicines
The author reflects on their journey with ancient medicines, sharing personal experiences that shaped their understanding of well-being. They describe a transformative period where they unplugged from the usual routines and embraced alternative healing practices. It’s not just about the medicines themselves, but the deeper connection to self and nature that these practices fostered. The author emphasizes the importance of balance and the lessons learned from both the highs and lows of this exploration. They highlight how these experiences have contributed to a more grounded and present version of themselves, encouraging others to consider their own paths to healing and self-discovery.
This Free Claude Skill Makes 100 Agents Fight So You Never See a Bad Response
You ask Claude for a launch plan. It gives you a decent first pass, except it forgets you only have six hours to pull it off. So you say, “Try again, but keep the six-hour limit.” The next plan fits the schedule. It also quietly drops the customer emails you asked for in the first message. By prompt five, you’re rereading your own chat to keep track of what Claude forgot. My first move here would be to save the useful bits and start a fresh session with a short brief. The goal, the constraints, and what a good answer has to include. One clean shot.
[Sponsor] WorkOS: Let Agents Use Your Tools Without Giving Them Free Rein
An agent’s job isn’t known until it runs, so a role assigned in advance is always too much or too little. Give the agent a task instead, and every call it makes gets checked against that intent, whether it’s reading a Linear issue or emailing a distribution list. With WorkOS Airlock, you set the policies once. Each call is allowed, denied, or sent to a human for approval, and every verdict is logged.
OS 162: Advanced/obtainable seating, no-name Italian forever pieces, anonymous glassware, rarest Pesce yet
Hello. Massive furniture selection on market this week. One of the last true weeks before Thanksgiving. A lot of premium items being offloaded. Biggest week since right after Labor Day. In market terms: there are really only a couple of weeks left in the year to sell and buy. This is one of them. This newsletter is straight to the action, all furniture, no anything else. Two/three dozen designer, high quality S-tier furniture items, listed with context and historical pricing for readers below.
Anthropic's super bug-finding Mythos model is going into Cisco and IBM products
Anthropic has started extending access to its bug-finding Mythos model by allowing Cisco and IBM to make it part of their security products. Cisco announced its integration today at an event in Melbourne, Australia, where it revealed that the red teaming tools in its AI Defense suite, which looks for weaknesses and inappropriate activity in AI models, agents, and applications they power, will be able to call Mythos to test customers’ setups.
Cisco to link all future products to its SaaS-y Cloud Control management tool
Cisco will no longer create products that don’t integrate with Cloud Control, its SaaS-y management platform that offers netadmins a single tool to manage all of the networking giant’s devices and much of its software portfolio. The networking giant announced Cloud Control in June 2025 and only made it available to customers in the USA. Cisco has previously told The Register that slow rollout reflects the fact that Cisco needs to nail privacy and other regulatory niceties, and deploy infrastructure, before introducing the service to other markets.
Lawyers Are Flying in Circles Above Nvidia’s Head After Jensen Huang’s Affair With Alleged Spy
Jensen Huang, the CEO of Nvidia, is facing significant legal challenges after admitting to an affair with Alice Huang, a top executive at a Singaporean firm, who is suspected of being a Chinese intelligence asset. This revelation has sparked multiple investigations, including one by the FBI, and has already led to a drop in Nvidia's stock value, erasing around $193 billion in market worth. Legal experts suggest that Huang’s position is precarious, with potential lawsuits alleging that Nvidia misled investors about the nature of the relationship and its implications for national security.
At least four law firms are now investigating Nvidia for possible securities fraud, and a class-action lawsuit has already been filed in California on behalf of an investor. The lawsuit cites misleading statements made by Nvidia's representatives regarding the relationship and its impact on the company’s dealings in China. The unusual corporate structure of Nvidia, where Huang holds significant control and lacks a clear succession plan, has raised further concerns about governance and oversight.
Experts are warning that if it turns out Huang was aware of Alice Huang's ties to Chinese intelligence, the fallout could be severe. While Nvidia's board has expressed confidence in Huang, the lack of a thorough investigation before making such statements could damage their credibility. The situation is evolving, and it’s clear that the implications for Huang and Nvidia could be far-reaching.