Cloudflare has unveiled a comprehensive internal AI engineering stack built entirely on its own platform. Ninety-three percent of the R&D team utilized AI coding tools powered by this infrastructure in the last month. The initiative, launched eleven months ago, now supports over 295 teams, generating 47.95 million requests and processing 241.37 billion tokens monthly. Developer velocity increased dramatically, with merge requests nearly doubling from a Q4 baseline.
To eliminate bottlenecks, Cloudflare deployed a sophisticated multi-agent AI code review system. Instead of a single model, up to seven distinct reviewers focus on security, performance, and compliance. A central coordinator agent manages these specialists, deduplicating findings and consolidating feedback. The company chose OpenCode for its server-first architecture and composable plugin design to ensure flexibility across tens of thousands of merge requests.
Google is assembling a strike team of researchers and engineers to improve its AI coding models and automate more of its own research. This push was prompted by Anthropic's recent AI breakthroughs, which raised the stakes in the race to develop advanced AI systems. The team includes some of Google's top AI researchers and focuses on generating and understanding code better, aiming to enhance Google's ability to automate research processes effectively.
In the realm of national security, the United States is accelerating new weapons design and procurement using AI to reduce development time from years to months. Autonomous technologies like the V-BAT drone can bypass enemy countermeasures while AI defends critical infrastructure in real-time against cyberattacks. Local decisions about AI infrastructure directly impact national security and technological edge as America and China race for supremacy.
Education faces ongoing challenges as the Purdue cheating scandal sparks debate over AI in classrooms. A professor accused many computer science students of using AI to cheat, causing over half the class to unenroll. Students criticized the professor for bypassing typical university procedures and fear-mongering without providing specific evidence. The university ultimately dropped the allegations, allowing students to re-enroll and continuing the broader campus conversation about AI's role.
Meanwhile, New Zealand is leveraging AI paired with geotechnical data to improve infrastructure through the upgraded NZGD database. Engineers can now navigate and query underground test data in natural language using the BEYON digital twin platform. The updated platform runs on Microsoft Azure and uses OpenAI's GPT-5.1 model for AI assistance, helping engineers advise clients on what can be built based on soil composition and rock layers.
Databricks will host its Data + AI Summit in San Francisco from June 15-18, 2026, focusing on AI agents and vibe coding. The event features over 20 expert-led courses including Vibe Coding on Databricks and Deploying and Monitoring Agents. Early registrants can secure a 50% discount on training which expires April 30, while a new Context Engineering certification exam launches for professionals validating expertise in agentic AI systems.
Harvey AI CEO Winston Weinberg discussed how the platform redefines white-collar jobs in the legal industry. The platform aims to speed up diligence processes and transform the economy in different ways by assisting professionals and automating tasks. The conversation highlights the potential impact of AI on traditional job roles and workflows, adding to the broader discourse on economic transformation.
Boehringer Ingelheim has opened an AI center for pharmaceutical research in London to expand its capabilities. The company will invest 150 million pounds over 10 years to develop targeted medicines for patients with unmet medical needs. The center is located near King's Cross station in the Knowledge Quarter area of London, marking the company's fourth location after Austria, Germany, and the U.S.
Yann LeCun, known as the Godfather of AI, criticized Anthropic CEO Dario Amodei for claiming AI will cause significant job displacement. LeCun stated that Amodei knows absolutely nothing about the effects of AI on jobs and called him wrong. LeCun emphasized that current AI technology is not advanced enough to replace human jobs on a large scale, highlighting the ongoing debate within the AI community about the future of work.
Key Takeaways
- Cloudflare's internal AI stack is used by 93% of its R&D team, processing 241.37 billion tokens monthly and doubling developer velocity.
- Cloudflare deployed a multi-agent code review system using up to seven distinct reviewers managed by a central coordinator agent.
- Google is forming a strike team to improve AI coding models and automate research, driven by Anthropic's recent breakthroughs.
- The U.S. is using AI to accelerate weapons design and procurement, reducing development time from years to months.
- A Purdue cheating scandal involving AI allegations has sparked a broader campus debate, though the university dropped the professor's claims.
- New Zealand's BEYON platform uses Microsoft Azure and OpenAI's GPT-5.1 to query geotechnical data for better infrastructure planning.
- Databricks is hosting a 2026 summit in San Francisco featuring courses on AI agents, vibe coding, and a new Context Engineering certification.
- Harvey AI aims to transform the legal industry by automating diligence processes and redefining white-collar job roles.
- Boehringer Ingelheim is investing 150 million pounds over 10 years in a London AI center for pharmaceutical research.
- Yann LeCun criticized Anthropic CEO Dario Amodei for underestimating the current limitations of AI's impact on the job market.
Cloudflare reveals internal AI engineering stack
Cloudflare unveiled its proprietary AI engineering stack built entirely on its own platform. Ninety-three percent of the company's R&D team used AI coding tools powered by this infrastructure in the last month. The initiative, launched eleven months ago, involved creating MCP servers and an access layer to integrate AI into workflows. Over 295 teams are now using these agentic AI tools, generating 47.95 million requests and processing 241.37 billion tokens monthly. Developer velocity increased dramatically with merge requests nearly doubling from a Q4 baseline.
Cloudflare deploys multi-agent AI code review system
Cloudflare engineered a sophisticated multi-agent AI code review system to eliminate bottlenecks and improve code quality. Instead of a single model, up to seven distinct reviewers focus on areas like security, performance, and compliance. A central coordinator agent manages these specialists, deduplicating findings and consolidating feedback into a single structured comment. This system has been tested across tens of thousands of merge requests with high accuracy. Cloudflare chose OpenCode for its server-first architecture and composable plugin design to ensure flexibility.
Cloudflare builds internal AI stack on shipping products
Cloudflare published a detailed breakdown of its internal AI engineering stack which runs on the same products it ships to customers. In the last 30 days, 93% of the R&D organization used AI coding tools powered by this infrastructure. The architecture spans three layers including the platform, knowledge, and enforcement layers. AI Gateway handles routing and cost tracking while Workers AI provides on-platform inference with open-weight models. The system supports over 3,600 internal users and 295 teams utilizing agentic AI tools.
Cloudflare unveils vision for agentic cloud infrastructure
Cloudflare officially unveiled its vision for the agentic cloud during Agents Week 2026 to support AI agents operating at scale. The company introduced new compute environments ranging from full operating systems to lightweight isolates. Artifacts offers versioned storage for code while Cloudflare Sandboxes provide persistent isolated environments for agents. Security features include Outbound Workers for zero-trust egress and Durable Objects for stateful code execution. The platform also includes primitives for inference, search, memory, and browser interaction.
Group essay strategy prevents AI cheating in college
A philosophy professor at the University of Chicago implemented a group essay assignment to prevent students from using AI for cheating. The professor replaced the role of ChatGPT by inviting students to collaborate on essays in class. This approach requires students to engage actively, defend their writing, and work together to shape the content. The strategy works particularly well in small classes where students must participate to understand the essay direction. This method shifts the focus from detecting AI to fostering genuine learning and collaboration.
Purdue cheating scandal sparks debate over AI in education
A Purdue professor accused many computer science students of using AI to cheat on assignments, causing over half the class to unenroll. The professor sent an email alleging AI use and threatening grades and disciplinary action for non-response. Students criticized the professor for bypassing typical university procedures and fear-mongering without providing specific evidence. The university ultimately dropped the allegations, allowing students to re-enroll in the course. The incident has sparked a broader campus conversation about the role of AI in education.
AI is the present not the future of national security
America and China are racing for technological supremacy with AI leadership being the key factor today. The United States is accelerating new weapons design and procurement using AI to reduce development time from years to months. Autonomous technologies like the V-BAT drone can bypass enemy countermeasures while AI defends critical infrastructure in real-time against cyberattacks. The U.S. needs to enable AI innovation to remain ahead of adversaries using AI-enabled attacks. Local decisions about AI infrastructure directly impact national security and technological edge.
Google forms strike team to improve coding models
Google assembled a strike team of researchers and engineers to improve its AI coding models and automate more of its own research. The push was prompted by Anthropic's recent AI breakthroughs which raised the stakes in the race to develop advanced AI systems. The team includes some of Google's top AI researchers and focuses on generating and understanding code better. They are also working on developing new AI models to automate more of their own research processes. The goal is to enhance Google's ability to generate and understand code effectively.
AI and geotechnical data improve New Zealand construction
New Zealand uses AI paired with geotechnical data to build better infrastructure through the upgraded NZGD database. Engineers can now navigate and query underground test data in natural language using the BEYON digital twin platform. The NZGD was established in 2013 after the Christchurch earthquake to centralize fragmented site data. The updated platform runs on Microsoft Azure and uses OpenAI's GPT-5.1 model for AI assistance. This system helps engineers advise clients on what can be built above ground based on soil composition and rock layers.
Databricks summit highlights AI agents and vibe coding
Databricks will host its Data + AI Summit in San Francisco from June 15-18, 2026, focusing on AI agents and vibe coding. The event features over 20 expert-led courses including Vibe Coding on Databricks and Deploying and Monitoring Agents. Early registrants can secure a 50% discount on training which expires April 30. A new Context Engineering certification exam is also launching for professionals validating expertise in agentic AI systems. The summit offers hands-on labs and personalized learning plans for attendees.
Harvey AI CEO discusses economic transformation
Harvey AI CEO Winston Weinberg discussed how the platform redefines white-collar jobs in the legal industry on The Claman Countdown. The platform aims to speed up diligence processes and transform the economy in different ways. Harvey AI is designed to assist professionals in the legal field by automating tasks and improving efficiency. The conversation highlights the potential impact of AI on traditional job roles and workflows.
Boehringer Ingelheim opens AI center in London
German drugmaker Boehringer Ingelheim launched an AI center for pharmaceutical research in London to expand its capabilities. The company will invest 150 million pounds over 10 years to develop targeted medicines for patients with unmet medical needs. The center is located near King's Cross station in the Knowledge Quarter area of London. This marks the company's fourth location after Austria, Germany, and the U.S. The UK government supports data-driven innovation in life sciences and healthcare.
It's not just one thing it's another thing
The sentence construction It's not just X it's Y has become so common in AI-generated writing that it serves as a reliable indicator of synthetic content. Barron's reports this pattern more than quadrupled from about 50 mentions in 2023 to over 200 uses in 2025. Generative AI tools were trained on human writing which includes this specific phrase structure. Em-dashes are also considered a tell for AI-generated text in addition to this sentence pattern. This trend reflects how reliant companies have become on AI for content creation.
Yann LeCun criticizes Dario Amodei on AI jobs
Yann LeCun, known as the Godfather of AI, criticized Anthropic CEO Dario Amodei for claiming AI will cause significant job displacement. LeCun stated that Amodei knows absolutely nothing about the effects of AI on jobs and called him wrong. LeCun emphasized that current AI technology is not advanced enough to replace human jobs on a large scale. His comments highlight the ongoing debate within the AI community about the future of work. The discussion underscores the complexity of predicting long-term effects of AI on the job market.
Sources
- Cloudflare's AI Engineering Stack Revealed
- Cloudflare's AI Code Review Overhaul
- The AI engineering stack we built internally — on the platform we ship
- Cloudflare Builds the Agentic Cloud
- Opinion | A group essay prevented my students from using AI
- ‘Cheating scandal’ in Purdue class sparks debate over AI’s place in college
- AI Isn’t the Future of Our National Security. It’s the Present.
- Google Creates Strike Team to Improve Coding Models
- Pairing geotechnical data with AI helps New Zealand to build better
- Databricks Summit: AI Agents, Vibe Coding Take Center Stage
- Harvey AI will speed up 'diligence' and transform the economy in different ways: Harvey AI CEO
- Boehringer Ingelheim launches AI centre for pharma research in London
- It's not just one thing — it's another thing
- Godfather of AI and Meta's most popular ex-employee Yann LeCun says Anthropic CEO Dario Amodei 'knows absolutely nothing' about AI effects on jobs: ‘Dario is wrong…’
Comments
Please log in to post a comment.