AI News Daily Briefing — February 11, 2026
Stripe's AI 'Minions' Now Merge 1,000+ Pull Requests Per Week
Stripe quietly revealed that its internal AI coding agents — dubbed Minions — are now autonomously generating and merging over 1,000 pull requests per week, all with human review in the loop. This isn't a research demo or a startup pitch deck projection. This is one of the world's most demanding production codebases, processing billions in payments, being meaningfully shaped by AI agents at scale.
The timing is notable. Just hours earlier, former GitHub CEO Thomas Dohmke announced a record $60M seed round at a $300M valuation for a new startup focused on managing AI-generated code. The implicit thesis: when agents write this much code this fast, the bottleneck shifts from generation to governance — review, quality, drift, and maintainability. Stripe's Minions validate the demand side of that bet.
For builders, the signal is unmistakable. The question is no longer whether AI agents will write production code, but how organizations will absorb, review, and maintain the output. If Stripe — with its legendary engineering culture — needs purpose-built tooling for this, every team shipping AI-assisted code will too. The infrastructure layer for agent-generated code is the next frontier.
AI Adoption Is Burning Out Its Biggest Champions
TechCrunch reports that employees who most enthusiastically adopted AI tools are now experiencing expanded workloads and blurred work-life boundaries. Productivity gains aren't freeing up time — they're raising the bar for output expectations, creating a new class of burnout.
Andrew Ng: AI Job Displacement Fears Overblown, Skills Demand Surging
Ng argues that actual job losses from AI remain limited, but demand for AI-literate workers is accelerating fast. The reframe matters: the labor market risk isn't mass unemployment, it's a widening skills gap between AI-fluent workers and everyone else.
Claude Code Agent Adds Multi-Agent Team Support
Anthropic's Claude Code now supports multi-agent configurations, letting developers orchestrate collaborative coding workflows. This moves Claude deeper into the agentic IDE space where Cursor and Windsurf already compete.
80% of Replies Are Now AI Bots, Says Levelsio
Developer Pieter Levels observes that roughly 80% of his social media replies are AI-generated. Combined with Trung Phan's callout of AI content mill tactics on X, the platform's signal-to-noise ratio is deteriorating fast — a real problem for builders who rely on social discourse.
WorldLabs Demonstrates Real-Time 4D Content Generation
WorldLabs is showing progress on 4D content creation with efficient transition mechanisms for real-time processing. Still early, but the move from static 3D generation to temporally coherent 4D is a significant technical step.
ElevenLabs Launches Expressive Voice Mode Across 70+ Languages
ElevenLabs' new Expressive Mode adds emotional nuance and tone control to AI voice agents, with coverage across 70+ languages including underserved dialects. The de-escalation and customer guidance capabilities point toward enterprise call center displacement.
Microsoft Explores Superconductors for AI Datacenter Power Density
Azure is investigating high-temperature superconductor technology to rewire datacenters, aiming to dramatically increase power density and efficiency for AI workloads. It's infrastructure moonshot thinking, but the energy wall for AI scaling is real enough to justify it.
Jony Ive's OpenAI Hardware Device Delayed to 2027
The AI device Jony Ive has been designing for OpenAI is pushed back and won't be called 'io' per a court filing. The delay underscores how hard it is to build novel AI hardware — even with infinite design pedigree and OpenAI's backing.
GPT 5.3 Codex Tops Vercel's Next.js AI Evaluation Leaderboard
OpenAI's latest Codex model leads Vercel's Next.js framework eval benchmark, reinforcing the trend of LLMs competing on framework-specific coding tasks. Theo's related commentary suggests this release is reshaping developer assumptions about AI-assisted development.
ThePrimeagen Ships ClawdBot Hardware Project Using Claude Tools
ThePrimeagen is building a USB-C hardware project in public using Claude-based tooling, a practical showcase of AI-assisted hardware development workflows gaining traction among developer influencers.
Jason Calacanis Pushes OpenClaw as Automation Play for Laid-Off Workers
Calacanis is calling on recently laid-off Amazon and Microsoft employees to learn OpenClaw and automate their former roles. Provocative framing aside, it signals growing investor interest in agent-based automation platforms.
Former GitHub CEO Raises Record $60M Seed for AI Code Management
Thomas Dohmke's new startup lands a $300M valuation to build tooling for managing AI-generated code. The thesis — that code governance becomes the bottleneck in an agent-heavy world — is directly validated by Stripe's Minions disclosure today.
Primary Ventures Closes Massive $625M Fund V for Seed Investing
The sheer size of this seed-focused fund signals that early-stage capital availability remains robust despite macro uncertainty. Seed rounds keep getting bigger — and more competitive.
Launch Capital Offers $25K to OpenClaw Founders
Jason Calacanis's fund is offering $25K plus mentorship to founders building on the OpenClaw automation platform — a fast bet on the agent-native startup ecosystem.
Apple Makes Four Promises to Developers on Fairer App Store Treatment
Responding to antitrust pressure, Apple commits to addressing developer complaints about market dominance. The promises are vague enough to be meaningless or meaningful — the details will matter more than the announcement.
JumpCloud Launches Venture Arm for Security Startups
Enterprise security platform JumpCloud establishes a VC division to invest in next-gen IT and security companies — another sign of platform companies using venture arms as strategic moats.
Mistral AI Announces Global 48-Hour Hackathon Across 8 Cities
Mistral partners with Weights & Biases and NVIDIA for a worldwide hackathon push — community-building investment that signals Mistral isn't ceding the developer ecosystem to OpenAI and Anthropic.
China-Backed Salt Typhoon Hackers Breach Singapore's Top 4 Telecoms
Singapore confirms state-sponsored Chinese hackers gained limited access to critical systems at the country's largest phone companies. Salt Typhoon's telecom targeting campaign continues to expand geographically — it's not just a US problem anymore.
Russia Throttles and Restricts Telegram Access
Russia escalates Telegram restrictions with speed throttling, a notable reversal for a platform that previously enjoyed relatively free access despite tensions with Moscow.
Discord Age Verification Signals the Gated Internet Era
Discord joins a growing wave of platforms implementing age verification, reshaping how accessibility and privacy work online. The trend is accelerating, and builders should assume identity verification will become a standard infrastructure requirement.
SpaceX Completes Cryoproof Tests on Super Heavy V3 Booster
Multi-day cryogenic testing validates the redesigned propellant systems and structural integrity of the next-gen Super Heavy booster. Starship's cadence continues to accelerate.
Google Expands Tools to Remove Sensitive Personal Data from Search
Google enhances user controls for requesting removal of private information and non-consensual explicit content from search results. A meaningful privacy improvement, though enforcement at scale remains the real test.
Autodesk Sues Google Over 'Flow' AI Video Tool Name
Trademark dispute over Google's generative video product name — a reminder that naming AI products in a crowded market is becoming a legal minefield.
Today's throughline is clear: AI agents are writing production code at industrial scale, and the ecosystem is scrambling to build the governance layer around it. Stripe's 1,000+ weekly agent-merged PRs, a record $60M seed for AI code management, and Claude's multi-agent support all point to the same conclusion — the bottleneck in software development is shifting from writing code to managing what agents produce. Builders should be thinking less about whether to adopt AI coding tools and more about the review, testing, and maintenance infrastructure that comes after.