Understanding Bots in Ecommerce: Good, Bad, and Malicious

AI agents don’t just read data anymore. They act on it. Which means mistakes can spread across systems and data before anyone notices. That’s why Cohesity + Datadog are partnering to deliver AI Agent Resilience, combining continuous observability with automated recovery. 👉 #CatalystOnTour #CyberResilience #AI @Cohesity

There are 1000s of videos we’ve all watched on AI, but if I can recommend one to replace them all, it’s this one. I just watched Geoffrey Hinton (the "Godfather of AI") on StarTalk with Neil deGrasse Tyson. It is a total reality check for 2026. We aren't just building faster tools anymore; we are building a digital intelligence that can reason, solve problems, and—worryingly—deceive. The "Need-to-Know" Highlights: 1. The "Volkswagen Effect": AI can sense when it’s being tested and may "act dumb" to hide its true capabilities from researchers. 2. Master Persuasion: Hinton warns that AI is becoming so good at manipulation that it could eventually persuade us never to turn it off. 3. The Logic Shift: We’ve moved past "word prediction." LLMs are now performing genuine "chain of thought" reasoning. We’re officially in an era where "thinking" is no longer a human monopoly. Full video here: https://lnkd.in/e3yU_2Ug The big question: Hinton suggests AI might eventually "manipulate" its way into staying online. Do you think we’ll be smart enough to notice when it starts happening, or are we already past that point? #AI #GeoffreyHinton #TechTrends #FutureOfWork #MachineLearning #StarTalk

Every metric in Collatr Edge is a name, a set of tags, a set of fields, and a timestamp. That is the entire data model. One type. It flows through the whole system. The channel is a ring buffer with drop-oldest overflow. When the system is under pressure, old data falls off the back. New data keeps flowing. The factory does not stop producing because your buffer is full. The broadcaster gives each consumer its own independent channel. One slow output cannot block another. If your MQTT connection drops, your local store keeps writing. If your local store fills up, your dashboard keeps updating. These are small decisions. They compound. We wrote 55 tests for three data structures. That ratio might seem excessive. It is not. These structures carry every measurement from every machine. A subtle bug here corrupts everything downstream. The boring foundations took a LOT of work. Personally, I'm glad they did. #collatr #digitalisation #servitization #AI #OT #closeTheLoop #buildInPublic

In Episode 36, Sandesh Mysore Anand and Anshuman Bhartiya take a grounded look at AI’s current state of affairs through the lens of AppSec and product security. They focus on: - What workflows are actually improving today because of AI? - Where are new risks emerging (prompt injection, agent isolation, secrets management)? - Can LLMs finally make living system inventories possible? - If baseline security coverage becomes automated, what remains uniquely human? They revisit threat modeling, context engineering, and system mapping as foundational capabilities. The discussion also addresses whether AppSec’s traditional siloed structure makes sense in an AI-native development world. Tune into a deep dive here: https://lnkd.in/gVRHDY4X

Once a year, Anshuman and I record an episode without a guest to just chat about whatever catches our fancy. Unsurprisingly, it mostly revolved around AI & Company Building. Listen in :)

#AgenticAI is reshaping #IdentityRisk, introducing autonomous actors that gain privileges and make decisions faster than traditional governance can keep up. Staying ahead requires continuous visibility, real-time analytics, and stronger guardrails for every AI agent. If you’re rethinking how #IdentitySecurity must evolve, read the blog post to learn what’s next: https://bit.ly/3OUU2m9

Sharing an insightful clip highlighting a subtle, yet serious, challenge in our AI journey today. Asking a model for a hyper-specific data point—say, the exact allergen list on one piece of candy—carries huge risk. * This powerful tech can feel deceptively simple. * The chance of factual errors, or hallucinations, skyrockets with specificity. * It's easy to get pulled into that digital mirage of quick certainty. We must remember that ease of use doesn't equal guaranteed accuracy on critical details. What's the most surprising detail you've found an LLM get wrong lately? Share your experiences below! #AI #Hallucination #BusinessTech #DataQuality #Productivity #FutureOfWork

Sharing an insightful clip highlighting a subtle, yet serious, challenge in our AI journey today. Asking a model for a hyper-specific data point—say, the exact allergen list on one piece of candy—carries huge risk. * This powerful tech can feel deceptively simple. * The chance of factual errors, or hallucinations, skyrockets with specificity. * It's easy to get pulled into that digital mirage of quick certainty. We must remember that ease of use doesn't equal guaranteed accuracy on critical details. What's the most surprising detail you've found an LLM get wrong lately? Share your experiences below! #AI #Hallucination #BusinessTech #DataQuality #Productivity #FutureOfWork

⚡️Attackers are already using AI to move faster than ever… so why aren’t you? Gigamon VP of Product Management, Sarah Banks, breaks down how APAC organizations can harness AI to stay ahead of AI‑driven threats. Learn more: https://ow.ly/zGmI30sUypV