AMD Stock Drops 4.5% as Perplexity Puts Agents on Local PCs

AMD's rollout of Perplexity's Portable Computer on supported Ryzen AI Max Windows systems brings AI agents directly onto users' local PCs, a move that immediately shifts how AI-driven traffic can originate from consumer hardware. By embedding sophisticated language models in the Ryzen AI ecosystem, Perplexity enables on‑device inference, meaning a surge of autonomous agents will now browse, scrape, and interact with websites without ever hitting cloud APIs. For site operators, this translates to a new class of high‑frequency, low‑latency bot traffic that mimics genuine human users but originates from personal machines, potentially inflating page‑view metrics and skewing analytics. Website owners should anticipate a spike in AI‑generated requests that bypass traditional bot‑detection signatures tied to known cloud endpoints. Monitoring for unusually consistent user‑agent strings, repeated access patterns from Ryzen‑powered devices, and elevated CPU‑intensive resource calls will be crucial. Updating your llms.txt to explicitly list “Ryzen AI Max” as a known AI client can help differentiate legitimate user traffic from Perplexity‑driven agents. **Actionable tips:** 1. Deploy real‑time device fingerprinting that flags the “Ryzen AI Max” hardware identifier and routes those sessions to a sandbox for behavior analysis. 2. Add a dedicated entry in your llms.txt, e.g., `User‑Agent: PerplexityPortable/1.0 (Ryzen AI Max)`, and configure your server to log and rate‑limit requests matching this pattern. 3. Use llms.txt’s `Disallow` directives to restrict access to high‑value endpoints (e.g., `/api/checkout`) for any AI agents unless they present a verified token, mitigating potential abuse from on‑device agents.