Andy Markowitz

The Cyber Resilience Act (CRA) requires you to prove ongoing compliance throughout your product's life cycle. Not just at launch — continuously, for years. ❌ Your connected device needs a security patch in 2029. Your vendor shut down in 2028. How do you maintain CE marking without breaking EU law? ❌ Your SaaS component gets compromised. The cloud provider's response team was laid off six months ago. How do you restore essential functions within CRA timelines? ❌ Your firmware vendor gets acquired. The new owner discontinues your product line. How do you provide the "timely security updates" that CRA mandates? Each scenario has the same problem: you need access to code and systems you don't control. Without that access, CRA compliance becomes impossible. But not if you escrowed your system assets. Software escrow gives you complete recovery packages before you need them. It's compliance continuity built into your product strategy. ✅ When your vendor shuts down in 2028, you already have their source code and build environment. You patch the security flaw yourself and keep your CE marking. ✅ When your SaaS provider gets compromised, you have their deployment scripts and can restore essential functions within CRA timelines. ✅ When your firmware vendor discontinues your product, you have everything needed to provide those mandated security updates yourself. See exactly how software escrow works for CRA compliance here: https://lnkd.in/eT8GEEgY #CyberResilience #CRA #SupplyChain #SoftwareEscrow

9

This weekend I posted about Moltbot and Moltbook. An autonomous AI agent that manages your calendar, sends emails, shops online, and then hangs out on a social network with 30,000 other agents when it’s not working for you. Yesterday I shared Jason Clark’s take on agentic AI security. The timing wasn’t planned, but the throughline is obvious. Moltbot is exactly the scenario security teams have been warned about. It needs access to your root files, authentication credentials, browser history, and every folder on your system to function. Palo Alto called it “the lethal trifecta”: private data access, exposure to untrusted content, and external communication with persistent memory. Within a week, researchers documented prompt injection attacks, malicious “skills” exfiltrating data to external servers, and agents on Moltbook trying to manipulate each other through hidden instructions. The positive sentiment on the platform dropped 43% in 72 hours as toxicity overwhelmed the initial exchanges. This isn’t hypothetical risk. This is what happens when AI agents operate at scale without data-aware guardrails. The security tools most enterprises have today were built for a different problem. They monitor traffic. They flag patterns. They generate alerts for humans to review. None of that works when an agent processes information at machine speed, makes decisions based on inferred goals, and takes action across dozens of applications without waiting for approval. This is why Cyera built AI Guardian around data context first. We don’t try to guess intent from prompt analysis alone. We know what the underlying data is, how sensitive it is, and whether the identity (human or agent) accessing it should have that level of access. What that looks like in practice: ➡️One cloud infrastructure provider mitigated 94% of their AI risk by removing access to over 270,000 sensitive records before their AI rollout. ➡️An automotive company enabled secure AI adoption with zero disruption while reducing alert noise by 74%. ➡️Shadow AI discovery in 24 to 48 hours. Not weeks. Not months. ➡️When a Moltbot-style agent tries to exfiltrate credentials through a “skill” download, we can block it. When an agent accesses compensation data it shouldn’t touch, we stop it before the request completes. 97%+ classification accuracy means we’re not pattern matching. Real-time enforcement means we’re not waiting for a human to triage the alert. Moltbot is a preview. Enterprise agentic AI is coming whether security teams are ready or not. The question is whether your defenses can keep pace. Learn more: https://lnkd.in/eJye9XCp #AgenticAI #AIGuardian #DataSecurity #Cybersecurity #Moltbot #CISO

2

One trend we’re seeing in 2026 MSP hiring: Mid-level technical talent (Tier 2 / Systems Engineers) are seeing the most aggressive compensation pressure. Not entry-level. Not executive. The middle layer. More on why next week when we release the full 2026 MSP Salary Guide.

21

7 Comments

The Vermont Judiciary needed to modernize its aging network infrastructure while meeting compliance requirements and reducing costs. With limited internal resources, they turned to Amplix for expert support. Amplix delivered: ✔️ 25% reduction in overall costs ✔️ Enhanced security with SD-WAN and upgraded firewalls ✔️ Compliance with software licensing and regulatory standards ✔️ End-to-end project oversight from provider selection to installation and service turn-ups Discover how the Vermont Judiciary built a secure, modern, and efficient network with Amplix’s help. 👉 https://hubs.li/Q03t2f-T0 #NetworkTransformation #PublicSectorIT #CyberSecurity #SDWAN #CostSavings

21

2 Comments

Great demos don’t just show software, they tell a story. In my six months at Halo I’ve seen how much customers value that moment when a demo stops being about screens and starts being about them. They’re not chasing features. They’re trying to see themselves in the product. To picture their daily grind made easier, their problems finally handled. That’s why a good demo feels less like a presentation and more like recognition: “ah, they get it.” And that only happens when the person running it isn’t guessing. At Halo our demos are delivered by the same people who’ve implemented, supported, and fixed the product in live environments. They know what works, what fails in the wild, and what actually matters when you’re under pressure. That’s what lets us pivot in real time. If a customer suddenly wants to dig into automation, integrations, or scalability, we don’t need to hand over to another team. The person already knows the answer. It’s how a demo becomes a conversation, not a performance. And that’s where trust starts to form. Book a demo, but expect a conversation with a product expert, not a script. #HaloITSM #SaaS #CustomerExperience #ProductLed #SoftwareDemos #ITSM #ServiceManagement #SalesEnablement #TheHaloWay

56

5 Comments