About
I'm here for product marketing takes, cybersecurity stakes, and jazz breaks.
Ask…
Articles by Ivan
-
Claude Code for Marketing: Can't Fake the Funk
Claude Code for Marketing: Can't Fake the Funk
I’m biased as a Product Marketer, but messaging has always been the bottleneck to scaling GTM. It doesn’t matter how…
17 Comments -
Takeaways from the 2024 Verizon DBIR
Takeaways from the 2024 Verizon DBIR
Among the highlights of any security professional’s annual calendar, the Verizon Data Breach Investigations Report…
2 Comments -
Take the 2021 Cloud IAM Survey
Take the 2021 Cloud IAM Survey
As a cloud professional, we know you have a lot of thoughts about IAM – good and bad, we get it! Take this brief survey…
-
Joining ScaleFT to Lead the BeyondCorp Movement
Joining ScaleFT to Lead the BeyondCorp Movement
I’m thrilled to announce that I’ve joined the ScaleFT team, where I just completed my first week being thrown to the…
-
Serverless - Let’s Cut Through the Noise
Serverless - Let’s Cut Through the Noise
On the surface, it appears that the Serverless buzz arrived overnight and already swept through the hype cycle in…
-
Why Iron.io Joined the Cloud Foundry Foundation
Why Iron.io Joined the Cloud Foundry Foundation
As of today, Iron.io formally joined the Cloud Foundry Foundation, the community behind the rapidly growing open source…
-
Event-Driven Computing with Iron.io & OpenShift
Event-Driven Computing with Iron.io & OpenShift
There's a theme we've been driving here at Iron.io that's beginning to resonate with developers and businesses alike –…
Activity
3K followers
-
Ivan Dwyer shared thisWhat the World Cup taught me about B2B SaaS. Look at the data, sure, but follow the vibes. Data can be misleading. The vibes are never wrong. 🇧🇷 😭
-
Ivan Dwyer shared thisData Gravity vs Workflow Gravity It’s the debate du jour. The real answer is, “why not both?” But let’s play this out for fun (and profit?). The argument for workflow gravity stems from the SaaS era being won by systems of record, but now in the AI era, the real value comes from agents doing the actual work. Once agents get deployed across the organization, the moat lies in their orchestration. The argument for data gravity stems from the fact that garbage in = garbage out, and it’s dangerous to apply automation at scale when the foundation is brittle. The moat lies in the underlying context. Both fair points. But if everything is getting rewritten in the AI era, it makes sense to apply an AI lens to both arguments. If agents truly live up to their promise - give a goal and a set of tools, they’ll figure it out - then where’s the gravity in the workflows? They get written in real time. And if the models keep getting better - window + reasoning, they’ll get to the answer - then where’s the gravity in the context? They'll assemble in real time. We're right back to “why not both?” But in a more deterministic sense. Workflow gravity is still with orchestration - but with the guardrails, governance, and guarantees that the enterprise demands. Data gravity is still with context - but in a trusted, verifiable state that the enterprise demands. This whole argument is one of many reasons why we at Axonius place the qualifier **decision-grade** on asset intelligence. The workflows and the context depend on it. PS - who is old enough to remember these commercials? Classic. 🍻
-
Ivan Dwyer shared thisI was fortunate enough to be a guest on Daniel Miessler's Unsupervised Learning podcast recently, where we talked about how AI is reshaping the asset management discipline into decision-grade asset intelligence (also AI). Great conversation, and great insights from Daniel, who has both lived this experience in the past and has as strong a view as anyone on what the future holds. Full episode in the comments below 👇
-
Ivan Dwyer shared thisAI is the Medium. MESSAGE(.)md is the Message. Raise your hand if you’re disappointed in AI writing output. Okay, now keep your hand up if you do it anyway. Great, same set of hands. Like everyone who cares about messaging quality and taste, I’ve been grappling with this for the past few years. Through my work building out a messaging system in Claude Code, I landed on a pattern. Introducing MESSAGE(.)md - a dynamic messaging spec for AI content factories. The structure is opinionated, the message uniquely yours. Use this as the context layer for your AI GTM. It’s designed for progressive loading, so your agents grab the right message for the right audience at the right time based on the task at hand - without breaking the token bank. Check out the links below for more details. You don't even have to comment CLAUDE 👇
-
Ivan Dwyer shared thisNow that Fable is out, we can just throw all of our security needs at it and we’re good, right? Cybersecurity is solved, hooray! Good news: nobody serious thinks this. The question about where AI can be safely applied to security remains top of mind. Another nugget from my interview with Chris H. a couple weeks ago: I see AI first being a strong recommendation engine for prioritization, and as it makes sense of the environment – learning behaviors, patterns, and decisions – it can take on more of the remediation work. This matters for how Security teams approach adopting AI in general, but more specifically around agent and harness design. I think it’s more likely that security agents will be a lot tighter in scope and guardrails than, say, coding agents. Design appropriately.
-
Ivan Dwyer shared thisThere’s quite the range of first impressions with Claude Fable. I understand the reaction from the security community, but also the position from Anthropic. When Mythos was first introduced a couple months ago, it demonstrated a step change in all capabilities across many domains, but notably in cybersecurity. Anthropic held the model back in the name of cybersecurity, so everyone naturally associated Mythos with cybersecurity. Mythos **today** is still a step change in all capabilities, still notably cybersecurity, so they continued to hold it back with the guardrails in Fable. But because we collectively associated Mythos with cybersecurity, the first thing we collectively did with Fable was try cybersecurity use cases - and got nerfed in the process by the hyper restrictive guardrails. Meanwhile every other domain is rocking out with the step change from Mythos. Annoying, but it is what it is. Frontier labs gonna frontier.
-
Ivan Dwyer shared thisNew models are fun, exciting, scary, and overwhelming at the same time. Beneath it all is your IT and Security teams enabling safe AI adoption - the trusted work horses that let the token machine go brrrrr without going bust (until Finance shows up). Today, Axonius announces an Anthropic adapter, connecting to Claude’s Compliance API to fetch organization assets to address Shadow AI and help reduce the attack surface. Blog with docs in the comments 👇
-
Ivan Dwyer shared thisI don’t usually wear hats, but gave a presentation to the field today around all things AI. It was internal, so can’t share what I said, but this hat came in handy. Thanks for the assist Abhishek Agrawal!
-
Ivan Dwyer shared thisReaders of Matt Levine have learned that everything is securities fraud. Now security teams are learning that everything is incident response. In a post-Mythos AI-era, the idea of periodic reconciliation is out the window. Everything must be continuous, everything must be dynamic. The good news is that IR teams wrote the playbook already, now is the time to bring that playbook to proactive and preventative security measures. One nugget from my Resilient Cyber interview with Chris H.. See the full episode in the comments below. 👇
-
Ivan Dwyer reacted on thisIvan Dwyer reacted on thisGood infographic of the harness/orchestration layer opportunity. Plenty of room to build great companies. courtesy Ramp
-
Ivan Dwyer liked thisIvan Dwyer liked thisWarning: This video contains actual humans behind the technology. We gathered the product and engineering teams in Tel Aviv recently for our Tech Day. It was an opportunity to connect and align on how we’re advancing the Axonius Asset Cloud across asset intelligence, exposure management, and AI security. Take a look inside the people and conversations shaping what comes next.
-
Ivan Dwyer reacted on thisI am incredibly excited about this collaboration with Claude for Teachers! At ASSISTments, we believe that true innovation in education doesn't mean replacing the teacher with an algorithm. It means putting the teacher squarely in the driver’s seat for their students’ learning, and giving every school system the data to support them at scale. After more than two decades of building research-backed math learning tools, we've arrived at a defining moment. We now know that AI can meaningfully help teachers do what they do best: teach. I am so proud of my team who worked so diligently to make this possible: Túlio Vieira de Souza, Jennifer Ramos, M.Ed, Hannah Pandolph, Ryan Emberling, Henry Aviles, Samara Bierria-Anderson, Angela Larkin (Mecca), and everyone who supported this effort. And huge thanks to the Anthropic team that collaborated with us!Ivan Dwyer reacted on thisWe're introducing Claude for Teachers: free access to premium Claude capabilities for verified K-12 educators in the US, with a library of teaching skills and a direct connection to evidence-based curricula, mapped to academic standards in all 50 states. Ask for a lesson plan, and Claude starts from your state standards and high-quality curricula by connecting through Learning Commons. It then drafts a plan and student-facing materials you can revise and take into class. Claude for Teachers is built for K-12 privacy. We never train our models on your conversations, and student information is protected by a data processing agreement written to comply with FERPA. Read more about how Claude for Teachers works: https://lnkd.in/eHdK8rw9
-
Ivan Dwyer reacted on thisIvan Dwyer reacted on thisSome news I'm late in sharing: after three and a half years, I've left Recorded Future. What a ride it was. We joined the Mastercard family. We built and scaled a great marketing team, rebranded the company (and won a a couple of awards doing it), repositioned the platform, and rebuilt the entire marketing stack including a new website. We ran our first brand campaign, brought Predict to New York for the first time, and somewhere in there I interviewed Colin Jost and Michael Che on stage. We helped elevate the category itself, with Gartner creating a Magic Quadrant for threat intelligence and Recorded Future named a Leader in the inaugural one. I traveled across the world to meet customers and colleagues. Thank you to Tom Wentworth for bringing me in, to my leadership team for taking the leap with me (Denise Lu David Lum Sarah Zimmel Amy Redell Angeline Mullinix), and to everyone else who supported me in the journey. I'm taking a real break for the first time in my career, and starting conversations about what's next. If you're building something interesting, let's talk.
-
Ivan Dwyer liked thisFor security and IT teams, few can answer one simple question: Who's using Claude in their environment? Watch Frederico Hakamine show how our Anthropic integration uncovers every Claude Enterprise user, pinpoints security risks, and turns visibility into action. If AI governance still feels like guesswork, this demo changes the conversation.Ivan Dwyer liked thisI recorded a walkthrough of our new Claude Anthropic integration. In 2 minutes, I get to... 1. [See Assets] I started with 1,900 users. The integration sees 600 in my Claude Enterprise. 2: [Find Risk Exposure: Shadow IT] Then I flipped the question to who’s using Claude Code (in their devices), but outside my enterprise subscription: I got 92 users. 3: [Act/Remediate without tickets] Then I mobilize Okta to make sure those users use Claude the right way. This same pattern works across 1,400+ systems, 36 asset types, no matter where things are hosted (on-prem, cloud, remote). That's why I'm betting my career in asset intelligence :)
-
Ivan Dwyer liked thisIvan Dwyer liked thisIn our testing with Anthropic Fable 5 last week we noticed something interesting: it did a better job than Opus in lowering the severity of a vulnerability. In the evolutionary arc of using AI models in cybersecurity, this matters just as much (perhaps more) as vulnerability discovery. With all the talk about how good “Mythos-class” models are at /finding/ vulnerabilities, as security professionals, we need to really be looking at how these models can help us /fix/. One big opportunity not getting much coverage is proper prioritization so that resources are invested in fixing the most important vulnerabilities. The opportunity cost of fixing the wrong thing cannot be understated. As an example, we have an internal facing codebase that we use for benchmarking models relative to their ability to help us improve the security of our code. Both Opus and Fable flagged the same known vulnerability, Opus labeled it as “high” while Fable correctly identified that exploiting the vulnerability required satisfying impossible pre-conditions and labeled it as “low”. In other words: Fable risk-ranked the vulnerability the same as our humans did while Opus inflated the severity. Had we handed this off to an engineering team, they would have spent time resolving a low priority vulnerability rather than on something that really mattered. In addition to the cost of having an engineering team focus on the wrong thing, for our Cybersecurity team, being handed a vulnerability triaged by the LLM along with a reasoned severity can more quickly apply their human judgement. If they start with "this is a high severity vulnerability" and then through their own investigation determine that it's non-exploitable, that's time they could have spent elsewhere. Instead, validating the reasoned explanation from the model they can quickly understand if it's something that need greater exploration. Again, they can focus on what matters. While the sample size is small, Fable is showing progress in the area that matters most right now: fixing vulnerabilities. Basically, all models are good at finding vulnerabilities to the point where you can basically pick any of the leading models for the task, but Fable illustrates that the next evolution is in helping engineers fix bugs in code.
-
Ivan Dwyer reacted on thisIvan Dwyer reacted on thisThis is what happens when you vibe code your fireworks system. Happy 4th! 🇺🇸
Experience
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Education
-
-
-
-
-
-
-
Publications
-
Servers? Where We're Going We Don't Need Servers.
DZone
See publicationArticle featured in the DZone Guide to Building and Deploying Applications on the Cloud
-
Get a Job, Container: A Serverless Workflow with Iron.io
DZone
See publicationGuest post for DZone.
-
Iron.io brings its job queuing, scheduling and processing tools to Cloud Foundry
Cloud Foundry
See publicationQ&A with the Cloud Foundry Foundation
Explore more posts
-
Leigh Marshall Dow
LaRel R. drops some terrific insight when it comes to government marketing. On Branding the Beltway, she breaks down the realities of selling to government buyers—compliance hurdles, the need for speed, and why storytelling + data + collaboration = winning. This convo is sharp, practical, and a must-listen for anyone in B2G. 🎙️ Listen: https://lnkd.in/ep7FmKqn 🍊 More episodes: https://lnkd.in/eG6E5E33
2 Comments -
Sammie Walker
One of the most actionable tips in Deepwatch’s new CISO 100 Days guide is simple but powerful: Executives don’t want dashboards. They want decisions. When reporting up: ✅ Translate alert tuning into reduced dwell time ✅ Tie detection improvements to uptime and customer trust ✅ Map telemetry to business systems, not log counts This is how security becomes a strategic function, not just a reactive one. 💡 Tip: Use “decision deltas” to anchor your board updates: What changed, why it mattered, and what action you took. #CISO
1 Comment -
Housna Hamadet
Detour Dog leveraged DNS's TXT record component to facilitate clandestine command delivery to breached websites in a bid to prompt scam redirections or illicit code retrieval and execution, a report from Infoblox Threat Intel researchers showed. Read more: https://gag.gl/dha61J , via SC Media
-
Christopher Jablonski
About three years ago, a colleague who was a in-residence CISO suggested we record a podcast about orgs needing an "AI czar." So we did. Since then, some companies have added CAIOs, Chief AI Architects, and the like to lead the charge into the GenAI and now, agentic AI world. For most, however, ensuring AI security and safety is a CISO concern. A new piece on Tanium Focal Point says CISOs need to double as "AI risk advisors" to educate the C-suite and boards on AI. “Even though AI is now one of the top three agenda items in nearly every board meeting I attend, most boards still don’t understand it,” said David Gee, a Sydney-based board risk adviser, former CISO for HSBC, and author of the new book A Day in the Life of a CISO. “They’re hiring directors with accounting and legal backgrounds, not people with experience in cyber, data, or AI.” Being an AI risk advisor is a tall order for a role that is already responsible for bridging all other technical aspects of IT with cybersecurity, overseeing incident response and operational resilience, aligning and contributing to the business strategy, and managing regulatory compliance. The largest global organizations may just want to deputize the CISO's second officer as the AI risk advisor. Whoever takes on the role could get a head start with 5 moves to become an AI risk adviser. Read now: https://lnkd.in/g4NKt74Y Orlando De Bruce Jacklyn Kellick Ellie Robson-Frisby Mia Wilcox Joann Doan Andrea Boston Timothy Mercer Joe Amodio
1 Comment
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content