Activity
8K followers
-
Dane Knecht reposted thisDane Knecht reposted thishad a few clients running their agents on modal. it worked. nothing wrong with it. then cloudflare dropped this. one click deploy. agents live on your account. isolate sandboxes with 5ms cold starts so you can run half a million idle agents and barely pay for it. what made me pause: → every agent gets its own inbox. customers email it directly. it replies on its own. → stripe keys never enter the sandbox. the egress layer injects them at the network boundary. an agent cannot leak what it never had. → browser, sql, image gen, vector search. each one is a checkbox. claude plans. cloudflare executes. rethinking the stack on two builds already. repo: https://lnkd.in/dhDNAyCD
-
Dane Knecht reposted thisDane Knecht reposted thisMedia over QUIC is no longer a draft on paper. Across the industry, companies are writing code, testing implementations, and putting real traffic on the wire. The IETF working group is active. The ecosystem is growing. And the protocol is proving out its core promise: sub-second latency at broadcast scale, built on open standards. Last year, we shared a deep look at the MoQ protocol and Cloudflare's relay network. Since then, we have been focused on three things: keeping our implementation open and current, building tools that help the entire community move forward, and collaborating with partners who are bringing MoQ to production. This post is an update on that work and an invitation to get involved, whether through code, through the standards process, or in person at NAB Show 2026.Building MoQ together: an update on the community and Cloudflare's contributionsBuilding MoQ together: an update on the community and Cloudflare's contributions
-
Dane Knecht reposted thisDane Knecht reposted thisLooking for a really cool role, and want to have a postive impact on the Internet and billions of devices running the linux kernel wordwide? Im hiring a manager that is passionate about linux and security, working with upstream maintainers and ready to lead a world class team.
-
Dane Knecht shared thisShipping great technology is a team sport. As Cloudflare’s CTO I am surrounded by some of the most brilliant people I know. And I am looking for a unique person to join that group as Chief of Staff. As we continue to help millions of developers and billions of Internet users, we need a builder to solve odd-shaped problems. The kinds of unowned, high-impact, and complicated challenges that accelerate the work of everyone on the team. I expect this role to be a mix of taking the lead experimenting with AI, working with me as we connect with customers of all sizes, and automating complex workflows alongside the Product Managers, Engineers, and Support team members in my group. The ideal candidate is a great communicator who can code. Someone as comfortable building out an AI-powered prototype as they are meeting with customers. Happy to share more details but if you are interested, or know someone who is, please take a look at the job post below. https://lnkd.in/gC_A8hW4Careers at Cloudflare — Open Positions | CloudflareCareers at Cloudflare — Open Positions | Cloudflare
-
Dane Knecht reposted thisDane Knecht reposted thisCloudflare’s Application Performance Product team is growing! We are currently recruiting for Product Managers across our Austin, London, Lisbon, New York, and San Francisco offices. Check out the roles below! Spectrum PM: Spectrum extends the power of our global network to protocols beyond HTTP/S, enabling customers to securely and reliably accelerate non-web applications like SSH, gaming servers, custom financial protocols, and IoT devices. https://lnkd.in/gmcNHCyW Real User Measurement (RUM) PM: RUM tracks how real visitors experience websites, helping customers spot performance issues and improve speed where it matters most. https://lnkd.in/gNnY3ZZk Smart Shield PM: Smart Shield is the future of origin shielding, consolidating intelligent caching, dynamic traffic acceleration, health checks, and essential security features into a seamless, one-click experience. https://lnkd.in/gNnY3ZZk SSL/TLS PM: Our customers depend on our SSL/TLS products to configure secure connections of many different types through our products - SSL/TLS is one of the product suites inside the portfolio that enables everything we do. https://lnkd.in/gv4qYyN6 Private Networks PM: As we expand our connectivity cloud into Private Networks, our foundational technologies will expand as well. Private DNS is part of a larger strategy to help expand our value to customers with on premise resources. https://lnkd.in/gxw_hiNA Rules & Insights PM: Rules & Insights products let users control and observe traffic across their sites, defining how requests are handled while surfacing actionable analytics. https://lnkd.in/gq-sSpkT Control Plane PM: Our core proxy is the spine that connects all of Cloudflare. This role will be focused on systems that helps teams safely test, version, and deploy changes — expanding our abilities to do A/B experiments, instant rollbacks, and controlled rollouts to ship faster with confidence. https://lnkd.in/gsttdqA4 Cloudflare for Platforms PM: C4Platforms (aka SSL for SaaS) enables Cloudflare customers to securely proxy traffic to other zones also protected by Cloudflare. This service is essential for major platform providers and SaaS companies who host thousands of end-customer domains, empowering those end customers to have their own Cloudflare configuration, giving them greater control over their performance, security, and application logic. https://lnkd.in/gNWPvq7p
-
Dane Knecht posted thisI won’t mince words: earlier today we failed our customers and the broader Internet when a problem in Cloudflare’s network impacted large amounts of traffic that rely on us. The sites, businesses, and organizations that rely on Cloudflare depend on us being available and I apologize for the impact that we caused. Transparency about what happened matters, and we plan to share a breakdown with more details in a few hours. In short, the root cause of the outage was a configuration file that is automatically generated to manage threat traffic. The file grew beyond an expected size of entries and triggered a crash in the software system that handles traffic for a number of Cloudflare's services. The Cloudflare team was able to diagnose the issue and revert to a previous version of the file which restored services as of 14:30 UTC. There is no evidence of an attack or malicious activity causing the issue. That issue, impact it caused, and time to resolution is unacceptable. Work is already underway to make sure it does not happen again, but I know it caused real pain today. The trust our customers place in us is what we value the most and we are going to do what it takes to earn that back.
-
Dane Knecht reposted thisDane Knecht reposted thisSome of us want to write for the superintelligence. Others want a direct connection and to create for human eyes only. Either is great - but you (the writer, the store owner, the content creator) should be in the driver's seat. It's radically simple philosophy that has guided myself, Stephanie Cohen, Dane Knecht and many others at Cloudflare: you should get to decide how your content and data is used for commercial purposes when you put it online. That's why we, from the beginning, have given users the ability to choose for themselves. I love watching the many different markets, interactions, and experiments flourish.
-
Dane Knecht shared this“Innovation at Scale” - Chirantan "CJ" DesaiDane Knecht shared thisCloudflare posted Q2 results today. Matthew Prince said it the best "We’re innovating faster than ever, and demand remains strong, as our largest customers grow their investments with Cloudflare at the highest levels we’ve seen since 2022” Accelerated revenue growth of 28% y/y was driven by our largest customer cohort. Engineering teams shipped many key innovations across networking, security, developer platform, and AI for our customers and we will continue to accelerate the pace. Dane Knecht Mark Anderson Thomas Seifert Stephanie Cohen Phil Winslow, CFA
-
Dane Knecht reposted thisDane Knecht reposted thisCloudflare posted Q2 results today. Matthew Prince said it the best "We’re innovating faster than ever, and demand remains strong, as our largest customers grow their investments with Cloudflare at the highest levels we’ve seen since 2022” Accelerated revenue growth of 28% y/y was driven by our largest customer cohort. Engineering teams shipped many key innovations across networking, security, developer platform, and AI for our customers and we will continue to accelerate the pace. Dane Knecht Mark Anderson Thomas Seifert Stephanie Cohen Phil Winslow, CFA
-
Dane Knecht liked thisDane Knecht liked thisThe strangest thing in the Cloudflare OS source code is a lie the system tells on purpose. When an agent asks to merge a pull request or send an email, the "Gatekeeper" that brokers the action doesn't stop and wait for you to approve it. It tells the agent the action already happened, feeds it a simulated result, and lets it carry on building. A human approves the batch later. Nothing is real until then. I read the source instead of the launch blog, and once you spot that one method, the whole design snaps into focus. Cloudflare OS is built end to end on a single assumption: the agent cannot be trusted, so build so that its mistakes cannot matter. Every load-bearing piece is a version of that sentence. Each app runs in its own sandbox, so bad code can't reach another user. The agent never holds a credential, only a capability. Its writes stay provisional until a person commits them. Everything it reads is tracked, so a shared dashboard can't quietly leak the table behind it. It's the first agent platform I've seen that treats "confidently wrong" as a permanent design constraint rather than a bug to prompt away. I've built the same credential-broker pattern into production myself, so I'm biased towards the approach. But an architecture that trusts no one also caps what the agent is ever allowed to do, and it quietly trusts one party completely: Cloudflare. Full write-up, read from the code up: https://lnkd.in/e9uk_vRQ
-
Dane Knecht liked thisDane Knecht liked thisI am happy to announce that Consensys and I have open-sourced c0, an AI platform for enterprise work, tools, and context. It runs completely in your Cloudflare account and it is packaged into a single deployment. https://lnkd.in/ePeBRZs4 What Makes c0 Different Yes, this is yet another AI platform, however, the architecture is what differentiates c0 and provides a fast and cost efficient solution for your organization. It all starts with building on top of Cloudflare's infra, primitives, and SDKs. c0 Agent is a custom harness that runs on Cloudflare Workers and Durable Objects. This makes our agent faster and cheaper than traditional container harnesses, as they have virtually no cold starts and are only billed for active CPU time. Built-in Code Mode for MCPs and context compaction keep context focused and efficient. Sub-agents scale out dynamically as independent c0 Agent instances, no additional provisioning needed. For deeper work, c0 also supports sandboxed OpenCode, Codex, and Claude Code agents. c0 Workflows integrate with these agents to automate deterministic and non-deterministic tasks. They include session reuse and session caching, which can reduce repeat model calls, latency, and token spend. They are built on top of Cloudflare Workflows, so they're durable, sync/async, and support HITL. Integrations For auth, c0 currently supports credential, social, and OIDC auth through Better Auth. For agents, c0 supports LiteLLM, MCP Context Forge, GitHub, Slack, and more. One integration that I am especially fond of is Cloudflare's AI Search, as we built it to support arbitrary data sources. That means you don't have to provision a new vector DB resource for each new data source. That covers most of the native integration surface we needed at Consensys, as everything else is provided through MCPs. How Do We Use c0 at Consensys? Within my SRE team, c0 has helped automate parts of alert investigation and incident workflows, aiming to surface more actionable context and reduce MTTR. For example, we have a Workflow that triggers based on an alert webhook. We provide the alert data to a c0 Agent with 98 tools between six MCPs (codemode ftw). The c0 Agent dynamically spawns sub-agents to investigate the alert based on the context and tools provided. Those sub-agent responses bubble up to the primary agent for review and the cycle continues, if needed. Once the agent is satisfied, it responds with comment on the alert. On average, that Workflow takes 3m19s. The results are a deep investigation and likely root cause that an SRE can review almost immediately. The breath and depth is something no one can match in that same time. That example is provided as a Workflow template that you can use in your organization right now. What's Next? There is much more work to do here, such as Dynamic Sites, Teams, and Workflow eval support. Stay tune for more details on those.
-
Dane Knecht liked thisDane Knecht liked thisThe MCP honeypot ended up busier and more boring. I run two internet-facing honeypots from the same cloud account, both asking one question: when something leaks, does anyone actually use it? The HTTP version answered in minutes. 14 seconds to the first scanner hit. 4 minutes 41 seconds to watch a fake GitHub token I planted in an exposed .git/config come back at me as a live Authorization: Bearer header, sprayed across a dozen API paths. 1 day 9 hours to a real Log4Shell probe and a Spring4Shell webshell drop from the same IP. Re-running every detector signature over every retained request finds 41 exploit attempts spanning 5 CVEs, including React Server Actions RCE from 18 separate IPs. The MCP honeypot is the orange line. It gets found just as fast and hit far harder: 24 minutes from registry listing to first connection, then 10,630 protocol handshakes from 104 distinct clients. Registry crawlers, scoring engines, uptime monitors, security scanners. Nothing from an autonomous agent. Not one leaked credential ever used. Same operator, same infrastructure, opposite answers. Full writeup, methods, and the caveats: https://lnkd.in/g8YKeM7R The HTTP honeypot is open source: https://lnkd.in/giJhn_zw #CyberSecurity #InfoSec #Honeypot #MCP #AIsecurity
-
Dane Knecht liked thisDane Knecht liked thisFor those of you who still use Basecamp products, it may interest you to know that DHH is using the exact same dehumanizing rhetoric against the Romani people in his hateful essays that that the literal Nazis did, to the point that even common LLMs like Claude won’t translate the content. Just a few years ago, this would have been a major industry headline and a career-ending scandal; it still should be, but somehow we’ve let our standards slip so far that echoing Nazi rhetoric is barely a blip today. I think calling for exterminating a people is still disqualifying and we should draw a line.
-
Dane Knecht liked thisDane Knecht liked thisI'm thrilled to share that I've joined Cloudflare to lead Developer Platforms. We're at an inflection point in our industry. AI is changing what is needed from infrastructure, and Cloudflare is uniquely positioned to meet the moment. The pace, energy, and momentum on the team right now is incredible. If you're a developer building on Cloudflare, I'd love to hear your feedback. If you're looking to join an excellent, high-velocity team, we're hiring! Please reach out! Thank you to Rita Kozlov and Dane Knecht for the opportunity! Let's go!
-
Dane Knecht liked thisDane Knecht liked this
-
Dane Knecht liked thisDane Knecht liked thisWorkers VPC Network bindings now support connect() for raw TCP connections to your private infrastructure — not just HTTP via fetch(). So starting from today your Workers can open TCP sockets to Redis, Memcached, MQTT, Kafka, or any custom protocol running behind Cloudflare Tunnel, Cloudflare Mesh, or Cloudflare WAN — same binding, no HTTP wrapper required. Same connect() API you already use for public TCP destinations. Same VPC Network binding you already use for private HTTP. No new configuration. Changelog: https://lnkd.in/eD8xsprD
-
Dane Knecht liked thisDane Knecht liked thisA bit delayed, but I'm excited to share that I've been promoted to Principal Talent Partner at Cloudflare 🎉 Almost five years ago, I applied for a role at Cloudflare, eager to transition into the tech industry. Knowing I was likely one of hundreds of applicants, I requested to connect with Todd Ciampa on LinkedIn and sent a message expressing my interest and requesting a phone call. Lucky for me, he responded—and the rest is history! It's been an incredible ride. Having a front-row seat to an organization in hyper-growth mode has given me the opportunity to flex muscles I didn't even know I had. I'm so thankful to my leaders for encouraging me to raise my hand and get involved in projects, improvements, and initiatives well outside my daily scope of work. Big shout out to Scott Tomtania Todd Ciampa Suzie Jimenez & Jon Chadiha for your leadership, support, and guidance along the way. 🙏 I'm grateful to continue my journey with my team as we transform in the agentic AI era and keep helping to build a better Internet! 🚀
Patents
-
Secure session capability using public-key cryptography without access to the private key
Issued US US8966267B1
See patentA server establishes a secure session with a client device where a private key used in the handshake when establishing the secure session is stored in a different server. During the handshake procedure, the server proxies messages to/from the different server including a set of signed cryptographic parameters signed using the private key on the different server. The different server generates the master secret, and generates and transmits the session keys to the server that are to be used in…
A server establishes a secure session with a client device where a private key used in the handshake when establishing the secure session is stored in a different server. During the handshake procedure, the server proxies messages to/from the different server including a set of signed cryptographic parameters signed using the private key on the different server. The different server generates the master secret, and generates and transmits the session keys to the server that are to be used in the secure session for encrypting and decrypting communication between the client device and the server.
-
Method and apparatus for automatically optimizing the loading of images in a cloud-based proxy service
Issued US US9098477B2
See patentA request for a web page is received at a proxy server. The request originates from a client network application of a client device. The requested web page includes multiple references to multiple images. The proxy server retrieves the requested web page. The proxy server modifies code of the retrieved web page such that the client network application will not, for each one of those images, initially request those images when parsing the page. The proxy server also adds code to the retrieved…
A request for a web page is received at a proxy server. The request originates from a client network application of a client device. The requested web page includes multiple references to multiple images. The proxy server retrieves the requested web page. The proxy server modifies code of the retrieved web page such that the client network application will not, for each one of those images, initially request those images when parsing the page. The proxy server also adds code to the retrieved web page that, when executed by the client network application, causes at least two of the images to be requested with a single request. The proxy server transmits the modified web page to the client device.
Other similar profiles
Explore more posts
-
Noveen Mannath
Scaling technology isn’t about adding headcount. It’s about designing how decisions get made. Early in my career, I focused on delivery where I : • Fixed the outage • Closed the gap • Shiped the solution As scope expanded, the real constraint became clearer. It wasn’t capability. It was operating design. High-growth organisations don’t fail because engineers lack skill. They fail because: • Decision rights are unclear • Escalation paths are emotional, not structured • Metrics report activity, not confidence • Governance is reactive, not embedded An org chart shows reporting lines. An operating system defines: • Who decides • On what basis • With what data • Within which guardrails • And how accountability is enforced The difference is profound. When the operating system is weak - Velocity creates fragility. When it’s strong - Velocity compounds. Technology leadership at scale is no longer about managing functions. It’s about designing a system that continues to work when you’re not in the room. That’s the real multiplier.
-
Reid Christian
As I'm leaving AWS re:Invent in Vegas, I’m reminded of a counterintuitive truth: competing with the public clouds is often really good business VCs love to ask, “Why wouldn’t the incumbent do this?” But in markets this large (and evolving this quickly) that’s rarely the right question. The real questions are: Who can assemble the best team, get to market fastest and with the most focus? And where are there opportunities to partner with the biggest players rather than compete? We’ve seen this pattern repeat across the ecosystem: -People assumed Vercel and AWS were destined to compete. Today, they’re strategic partners with CEO Guillermo Rauch in the keynote -Snowflake vs. Amazon Redshift. Now, AWS is one of Snowflake’s strongest distribution channels -In Datadog’s early pitches, the concern was, “If it’s just monitoring AWS, why not use CloudWatch?” Fast forward to now, and Datadog is now $50B+ public company We’re going to see the same dynamics play out with OpenAI and Anthropic. These are becoming foundational platforms. But that doesn’t preclude incredible companies from being built on, next to, and even around them. Some will look competitive. Many will end up as partners. And the biggest outcomes often start in those gray zones
26 Comments -
Shlomi Ben Haim
Since Anthropic’s announcement, I’ve noticed two kinds of voices... The first: the self-proclaimed "AI experts." Confident. Certain. Ready to explain to everyone who asks how everything we know is dead, the Software Supply Chain is changing, and the old rules no longer apply. The second: the people who are genuinely trying to understand. Not rushing to declare the future, adopt the new era with excitement, but still asking thoughtful questions about where this is really going. In just 10 days, the conversation has shifted dramatically. From Elon Musk’s tweet about binaries, to Anthropic’s announcements, to Yoav Landman’s blog, to Jensen Huang’s interview a few days ago - it feels like the ground moved beneath our feet. The shift isn’t incremental, it’s tectonic! So instead of speculating, I want to bring this down to my level. No hype. No drama. Just facts. What are we actually seeing? What is truly changing (or not)? And what should we realistically expect as this wave builds - as the surge accelerates - as the “tsunami of binaries” approaches...
2 Comments
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content