{"slug": "from-one-desk-to-the-whole-enterprise-making-local-ai-resilient", "title": "From One Desk to the Whole Enterprise: Making Local AI Resilient", "summary": "Cisco and AMD are partnering to build an enterprise-ready architecture for local AI on AMD Ryzen AI Halo hardware, addressing network, tokenomics, agent behavior, and security challenges. The solution integrates Cisco's observability, governance, and security tools to turn local AI from a standalone device into a managed, scalable enterprise system.", "body_md": "This is a breakthrough moment for enterprise AI. Leaner models, an open and ready software stack, and powerful hardware like AMD Ryzen™ AI Halo now make it possible to run serious AI right at the desk. Together, these advances mean enterprises can run serious AI where their people work, not just in a distant cloud. The hardest part is still ahead: turning what works on one developer’s desk into something thousands of users across an enterprise can rely on, securely and at scale.\n\nThat’s why, together with AMD, we’re building AI resilience for AMD’s Ryzen™ AI Halo, a solution and set of integrations that turns local AI from a standalone device into an enterprise-ready architecture.\n\nAbove: The four first-class problems for deskside AI at enterprise scale\n\n**The shift no one can ignore**\n\nAs enterprise AI moves from experimentation to deployment, agentic inference is reshaping infrastructure from the ground up. We’re going from bursts of traffic from human-led chatbots to agents that run 24/7 and never sleep, generating 450% more network traffic than human operators doing the same work. Humans click; agents swarm.\n\nThe need for token efficiency and data sovereignty is driving a new class of computing, deskside computing, with users and teams putting AI agents right by their sides. Inference is moving to a hybrid architecture with thousands of ambient deskside agents in an enterprise helping employees have 24×7 productivity. That’s an extraordinary opportunity. It’s also a brand-new operating challenge.\n\n**But, you can’t just put a powerful machine on every desk and hope for the best**\n\nTo make deskside and local AI computing work at enterprise scale, every AI node must be treated as a secure, managed node in the enterprise network. When we look at what enterprises actually need to get there, four first-class problems emerge:\n\n**Network**– the fabric has to handle the agent traffic without buckling while ensuring only the necessary network access is provisioned.**Tokenomics**– ensure we take advantage of utilizing local inference to limit token costs from frontier LLMs.** Agent behavior**– observe and enforce what deskside agents can & cannot do.** Security**– this new operating model leads to new threats & vulnerabilities which must be actively managed.\n\nThese aren’t afterthoughts. They’re the foundation. And they’re exactly where AMD and Cisco are partnering to deliver.\n\n**A partnership that turns local AI into an enterprise architecture**\n\nAMD provides the deskside / local AI platform. At the foundation is AMD Ryzen™ AI Halo hardware, an isolated agent sandbox and the services needed for local-first inferencing, including model routing and token limits via AMD’s Semantic Router and local inference on Lemonade.\n\nCisco wraps that platform in a secure harness—the observability, governance, and control enterprises need, all in one seamless experience:\n\n**Splunk Agent Observability + Splunk Infrastructure Monitoring** provides a fleet-wide full-stack observability tracking agent behavior, tokenomics and compute utilization.**AI Defense** for model and agent security.**DefenseClaw** for security policy enforcement, so guardrails are enforced directly on-device, within the agent harness.**Cisco Cloud Control** as the single pane of glass for unified policy and control.\n\nTogether, spanning the Cisco Secure Network beneath it all, this transforms local AI into an enterprise-ready architecture, not a standalone device.\n\n**What it looks like in action**\n\nThrough Cisco Cloud Control, IT teams gain the operating layer around their entire Ryzen™ AI Halo fleet:\n\n**See everything:** Correlate each Ryzen AI™ Halo device with its employees, its agent identities, its security posture, and its Cisco network identity—all in one view. Drill down into utilization, throughput, and energy consumption, right down to individual agents running on a single device.**Optimize the economics:** A tokenomics view shows how AI work is split between AMD local execution on Lemonade inference and frontier providers, translates that into cost savings from AMD’s Semantic Router, and even highlights cloud workloads that could move onto Ryzen AI™ Halo devices for better economics.**Govern agent behavior:** Enforce a holistic set of guardrails, from unapproved usage patterns to risky agent actions, including deletion controls that restrict file access and tool calls before damage is done.**Contain what goes wrong:** When there’s a critical trust failure with an agent or model, use the Cisco network itself to place the offender in complete quarantine for investigation, and notify the owner. This is the differentiator: control that extends beyond the box, into the network.\n\nAbove: Tokenomics Insights\n\nAbove: Cisco’s Identity Service Engine AMD Ryzen™ AI Halo\n\n**The organizations that will win**\n\nA core principle behind our partnership with AMD is openness — giving customers the freedom to choose the models, frameworks, and deployment environments that fit their needs. But openness alone isn’t the finish line.\n\nThe organizations that truly succeed with AI won’t be the ones that simply adopt it. They’ll be the ones that can deploy it everywhere, see it clearly, govern it confidently, and control it decisively. Building the stack with the necessary resilience, without slowing their people down. That’s the promise of deskside AI, and it’s what our AI resilience solution for AMD Ryzen™ AI Halo is built to deliver.\n\nThe deskside AI era is here. Let’s make it resilient.", "url": "https://wpnews.pro/news/from-one-desk-to-the-whole-enterprise-making-local-ai-resilient", "canonical_source": "https://blogs.cisco.com/ai/from-one-desk-to-the-whole-enterprise-making-local-ai-resilient", "published_at": "2026-07-23 18:03:22+00:00", "updated_at": "2026-07-23 18:11:33.762623+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-infrastructure", "ai-agents", "ai-safety", "ai-policy"], "entities": ["Cisco", "AMD", "AMD Ryzen AI Halo", "Splunk", "DefenseClaw", "Cisco Cloud Control", "AI Defense", "Lemonade"], "alternates": {"html": "https://wpnews.pro/news/from-one-desk-to-the-whole-enterprise-making-local-ai-resilient", "markdown": "https://wpnews.pro/news/from-one-desk-to-the-whole-enterprise-making-local-ai-resilient.md", "text": "https://wpnews.pro/news/from-one-desk-to-the-whole-enterprise-making-local-ai-resilient.txt", "jsonld": "https://wpnews.pro/news/from-one-desk-to-the-whole-enterprise-making-local-ai-resilient.jsonld"}}