Auto-Arch Tournament vs OpenSRE

Auto-Arch Tournament is an autonomous research system where an AI agent iteratively proposes, implements, and validates microarchitectural improvements to a RISC-V CPU. Starting from a standard 5-stage pipeline, the loop runs hypotheses in parallel, each going through formal verification (53 symbolic checks), cycle-accurate simulation, multi-seed FPGA place-and-route, and CoreMark CRC validation. Only hypotheses that beat the current champion get merged; everything else gets discarded. Starting from 301 iterations/second, the system hit 577 iter/s (+92%) across 73 attempts in 9.8 hours — producing a design 26% faster and 40% smaller in LUTs than the baseline. The insight the author drives home is that the real innovation isn't the AI agent — it's the verifier. The orchestrator is hardcoded to prevent agents from manipulating their own evaluation gates, a simple but critical design constraint that turns a creative process into a trustworthy one. Without a rigorous verification harness, agent-driven optimization becomes a confidence trick. This is early but fascinating proof that AI-driven hardware design loops can produce commercially meaningful gains. The repo uses Claude Code or Codex as the coding agent, SystemVerilog for the RTL, and standard open-source EDA tooling (Yosys, nextpnr, Verilator). It's a compelling template for anyone building agentic optimization loops where correctness matters.

OpenSRE is an open-source toolkit from Tracer-Cloud for building AI-powered Site Reliability Engineering agents that can autonomously investigate production incidents. It connects to 40+ observability and infrastructure tools — logs, metrics, traces, runbooks, Kubernetes events, PagerDuty alerts — and uses parallel hypothesis testing to correlate signals across the stack without waiting for human direction. The agent follows a structured investigation protocol: it ingests the alert, builds a set of possible root causes, tests each hypothesis by querying the appropriate data sources, ranks them by confidence, and outputs a remediation plan with evidence attached. If configured, it can also apply low-risk fixes (e.g., restarting a pod, scaling a deployment) automatically and page the human only when it needs approval for higher-risk changes. Supports Anthropic Claude, OpenAI GPT, and local Ollama backends. The project sits at 1,250+ GitHub stars with a public beta available now. It fills a real gap in the open-source observability stack — while Azure SRE Agent and similar proprietary tools exist, OpenSRE is the first production-ready OSS option. The Tracer-Cloud team has been building production tracing infrastructure for three years and designed OpenSRE around actual on-call workflows.