ToolsSupabase adapts infrastructure for AI-driven surge in applicationsSupabase co-founders Paul Copplestone and Ant Wilson discussed how the company is redesigning backend infrastructure to support AI-generated applications.3 min read
ToolsNVIDIA Minerva correlates telemetry to reduce GPU downtime in large AI data centersNVIDIA Minerva is an AI-native AIOps platform that consolidates telemetry from GPUs, networking, storage and schedulers into a dynamic topology-aware knowledge graph.6 min read
ToolsNVIDIA Mission Control 2.5 brings self-healing recovery to Kubernetes AI workloadsNVIDIA has extended the autonomous recovery engine in Mission Control 2.5 to support Kubernetes-based AI workloads managed with NVIDIA Run:ai, in addition to its existing Slurm support for HPC.5 min read
ToolsCodeRabbit uses GPT‑6.1 Sol to surface security risks across large codebasesCodeRabbit has integrated GPT‑6.1 Sol as the main model in its security agent to analyze pull requests and investigate vulnerabilities across repositories, dependencies and shared components.2 min read
ToolsWarning: Do not add multiple accounts to T3 Code — CLIProxy (prox) shares subscriptions across machinesA user warns against placing multiple accounts in T3 Code and presents their fork of CLIProxy, called “prox,” which shares 7 Claude and 3 Codex subscriptions across 12 computers; its advantage is…1 min read
ToolsOpenAI engineers speed incident response using Codex's Ultrafast modeOpenAI engineers report that Ultrafast, a high-speed mode in Codex, has significantly reduced the time needed for incident and security investigations, enabling near-real-time answers and parallel searches across data sources.3 min read
ToolsMicrosoft's MAI-DxO system doubles diagnostic accuracy in complex simulated casesMicrosoft tested its Microsoft AI Diagnostic Orchestrator (MAI-DxO) in simulation studies published in Science and found it diagnosed very difficult cases correctly more than twice as often as general physicians using reference materials.4 min read
ToolsOpenAI engineer Rennsong collects feedback and fixes bugs with a 'dot' toolRennsong, an OpenAI engineer, uses his own 'dot' tool to gather social media and user reports, update a live tracker in the ChatGPT Space, and investigate and fix bugs in Codex.1 min read
ToolsMozilla CTO Raffi Krikorian: build open, personal AI infrastructureRaffi Krikorian, Chief Technology Officer of Mozilla, argued at AI Codecon that open-source models must be surrounded by interoperable infrastructure — memory, tool-calling, runtimes and guardrails — to deliver personal, sovereign AI.5 min read
ToolsTensorRT's Global Performance Tuner: model-specific build-route tuning for end-to-end performanceNVIDIA introduced the Global Performance Tuner in TensorRT 11.1 to enable automatic, accuracy-aware, end-to-end performance tuning by exploring model-specific build routes.5 min read
ToolsHow OpenAI Scaled Habitat’s Azure Cosmos DB Layer to Serve Global DemandOpenAI describes how it evolved Habitat—the company’s online storage platform—from a simple client library into a globally distributed storage layer supporting hundreds of petabytes and millions of requests per second.7 min read
ToolsProof-of-human tech lets people link verified humans to AI agents, reshaping online commerceStartups are developing proof-of-human systems that tie a verified person to AI agents acting on their behalf, potentially allowing merchants to distinguish trusted agents from bots without blanket bans.2 min read
