Good morning {{first_name| Artificial intelligence lovers}}. Catching up to Anthropic’s leadership in programming has become a personal project for Sergey Brin.
The Google co-founder, who was retired, is now said to be running a DeepMind “attack team” focused on closing Gemini’s internal programming gap with Claude – with Brin framing the push as the shortest path to the Holy Grail of self-improvement AI.
-
Brain mobilizes DeepMind to hunt down the Anthropists for the code
-
Moonshot AI’s Kimi K2.6 bridges the open source gap
-
Create high-converting landing pages in Claude Design
-
Adobe’s new agent AI platform for enterprise
-
4 new tools for AI, community workflow, and more
Image source: Louvart/Rundown
Rundown: Google co-founder Sergey Brin in person crowd DeepMind is outpacing Anthropic with Gemini, according to The Information, creating a new “attack team” and directing efforts to employees as the shortest path to self-improving AI systems.
-
Research engineer Sébastien Bourgoud, who previously ran pre-training at DeepMind, leads the group under CTO Koray Kavukcuoglu and Brin.
-
In an internal memo, Brain told employees that the real prize is the AI that trains the next AI, with programming being the ability that gets Gemini there.
-
DeepMind researchers are said to rate Claude’s code writing higher than Gemini’s internally, which led to Brain paying a dedicated team.
-
Gemini engineers now have to use Google’s internal agent tools for complex tasks, tracking usage on a company leaderboard called Jetski.
Why it matters: After dominating the AI conversation at the end of 2025, Google is off to a slow start into 2026. But Brain’s push isn’t a product response, it’s an internal response, and the attack team’s real mission is to automate Google itself, bridging the gap with the deeply embedded AI systems already running inside Anthropic and OpenAI.
Together with YOU.COM
Rundown: Most teams choose an API by checking a benchmark table and calling it “Implemented,” an abbreviation that may miss what really matters in production. This guide is from Ant.com Explains why initial response time is a misleading signal and what should be measured instead.
-
Why does p50 latency mask the failures users are actually experiencing
-
The “time to useful result” framework captures what the standards ignore
-
Four hidden cost drivers that show up in your records, not vendors’ spreadsheets
-
How to evaluate APIs on your actual concurrency levels, not on experimental conditions
Moonshot Amnesty International
Image source: Moonshot AI
Rundown: Kimi Moonshot AI Open source K2.6, a new proxy coding model that comes close to or exceeds models like GPT-5.4, Opus 4.6, and Gemini 3.1 Pro across the highest standards of inference, encryption, and more at a fraction of the cost.
-
K2.6 outperforms GPT-5.4, Opus 4.6, and Gemini 3.1 Pro in benchmarks including Humanity’s Last Exam with Tools (inference) and SWE-Bench Pro (coding).
-
In long-term work, the K2.6 can run for over 12 hours straight across over 4,000 tool calls, with Kimi showing it rebuilding an 8-year-old code base in demos.
-
Perpetual agents like OpenClaw and Hermes are now running on K2.6, with Kimi reporting one internal agent running autonomously for five days in a row.
-
K2.6 agent swarms can now spin up to 300 parallel sub-agents at the same time to complete tasks, three times the number of its K2.5 predecessor.
Why it matters: Dario Amodei just said that open source and China are 6-12 months behind frontier labs, and while this may be true for internal releases, public systems seem much closer. Given the frustrations with usage rates and the rise of independent agents, K2.6 looks like a powerful and cost-effective new option for agent workflows.
Artificial intelligence training
Rundown: In this guide, you’ll learn how to use Claude Design, Anthropic’s new design tool, to create four different landing page variations for your website.
-
Go to claude.ai/design, select the wireframe option, click Create, and write: Who is the page aimed at, what are you selling, and what action should the visitor take?
-
Find a landing page you like and shoot it. Try searching for “top landing pages in [your niche]“The payment part of a site that does millions of daily transactions, such as Amazon, eBay, or PayPal
-
Add screenshots to your summary, and ask Claude to create four different variations of the mockup. Answer any follow-ups, and wait 2-5 minutes for results
-
Click on any item and leave a note like “Rewrite CTA to be results-specific” or “Add testimonial.” Claude applies changes to improve output
Pro Tip: Click Share > Deliver to Claude Code > Submit to Claude Code Web to get Claude Code to create and publish the final website for you.
Provided by Slack
Rundown: Your team is already using AI tools. But are they actually getting smarter results? Slackbot, the AI built into Slack, turns your entire workspace into a productivity engine-surfacing answers, crafting content, and automating busy work so your team can focus on what matters.
In this free guide, you’ll discover:
-
How Slackbot aggregates conversations, files, and apps for instant answers
-
Ways to automate routine tasks without writing a single line of code
-
How teams use AI to move faster and collaborate more effectively
Adobe
Rundown: Adobe only foot CX Enterprise at Adobe Summit, a new agent platform designed to help companies orchestrate marketing, content, and customer interactions through AI agent networks.
-
CX Enterprise weaves three pillars under a single agent orchestration layer: brand visibility, content supply chain, and customer engagement.
-
CX Enterprise Coworker assembles the right agents and tools based on a specific user goal, creates a plan and executes multi-step actions.
-
The Adobe Marketing Agent is now connected to systems such as ChatGPT, Claude, Gemini, and Copilot, to coordinate between agents and Adobe applications.
-
The company is also launching an agent skills catalog, enabling organizations to create reusable and customizable workflows within the platform.
Why it matters: The entire design world is moving towards agent workflows, with Figma Agents, Canva Agents, and Adobe all vying for position. The biggest threat is the parameterization of the middleman: launches like Claude Design and each subsequent improvement will make old formatting paths more difficult.
-
🤖 Scrunch – Learn how AI interprets your website, performs a free audit, and unlocks the new way to reach customers*
-
⚙️ Kimi K2.6 – Moonshot AI’s powerful open source agent and coding model
-
🧠 Chronicle – OpenAI’s Codex feature that uses screen context for memory
-
🚀 Qwen3.6-Plus – Alibaba’s new AI with 1M context window, and powerful coding
OpenAI Rolled Chronicle, a Codex preview feature that runs background agents that capture your screen to build lasting memories, is initially limited to power users on Macs.
Former Meta Chief Science Officer for Artificial Intelligence Yan to be He said People shouldn’t listen to Dario Amodei about the impact of AI on labor markets, or “Sam, or Joshua, or Jeff, or me,” saying that economists have the more important perspective.
lovable to reject It reportedly suffered a data breach after users reported that public project chats were visible, saying the issue was an authentication failure.
Tinder and Zoom a partner With the world of Sam Altman, allowing users to earn “Proof of Humanity” badges via iris scans to combat AI bots and deepfakes.
Anthropic Expanded Amazon’s own deal for 5 gigawatts in computing, with the tech giant investing up to an additional $25 billion in Anthropic versus its $100 billion+ AWS commitment.
Recursive superintelligence He grew up $500 million at a $4 billion valuation, with a four-month-old startup founded by OAI and Deepmind alumni to build artificial intelligence that improves itself.
In each newsletter we show how the reader is using AI to work smarter, save time or make life easier.
Today’s workflow comes from the reader Julie N. In Grand Junction, Colorado:
“As a digital marketer, I use ChatGPT as my assistant. I will input all my information about a client’s Google Ads needs, including budget and landing page, and allow them to design keyword-friendly titles and descriptions, set the ad strategy, and recommend additional options for the campaign.
Also, I will take my ideas for a potential SEO package and create a PDF of the front-end and administrative side that I can share with potential clients.
How do you use artificial intelligence? Tell us here.
That’s all for today!
Before you go, we’d love to know what you thought of today’s newsletter to help us improve The Rundown experience for you.
Rowan, Joey, Zack, Shubham and Jennifer – the humans behind The Rundown
