{"company":{"name":"Komodor","slug":"komodor","website":"https://komodor.com","category":"developer-tools"},"question":"What has Komodor shipped recently?","answer":"In the last 30 days, Komodor shipped 10 tracked updates. The most recent was \"AI Agent Observability: What to Track on Every Agent Run\" on 2026-11-11.","window":{"days":30,"updateCount":10,"returned":10},"generatedAt":"2026-11-22T11:33:03.426Z","updates":[{"title":"AI Agent Observability: What to Track on Every Agent Run","summary":"Komodor introduced AI agent observability to reconstruct and explain every agent run, tracking triggers, model/tool calls, timing, cost, and outcomes. Unlike standard monitoring, it captures the full chain of actions in agent workflows, including cost per run as a key operational metric. The feature is built into Komodor’s Agentic Operations Platform and exports traces via OpenTelemetry to existing observability stacks.","date":"2026-11-11","dateIsEstimated":false,"signalType":"feature_update","signalTypeLabel":"Feature","sourceUrl":"https://komodor.com/blog/ai-agent-observability/","publisher":"komodor.com"},{"title":"Building AI SRE Agents, Part 4: One Agent Is a Tool. A Fleet Is a System.","summary":"Komodor’s latest blog post outlines five critical challenges teams face when scaling AI SRE agent fleets beyond a single agent, including silent quality drift, guardrail erosion, prompt injection risks, hypothesis storms, and unmanaged changes. It provides monitoring, mitigation, and guardrail strategies for each, emphasizing structural guardrails, version pinning, and coordination patterns.","date":"2026-11-11","dateIsEstimated":false,"signalType":"content_marketing","signalTypeLabel":"Content","sourceUrl":"https://komodor.com/blog/building-ai-sre-agents-part-4-agent-fleet/","publisher":"komodor.com"},{"title":"Komodor adds agent import capability for Google ADK, LangChain, Claude Agent SDK, and Agno","summary":"Komodor introduced a feature allowing users to import existing agents (Google ADK, LangChain, Claude Agent SDK, or Agno) into its Agentic Operation Platform (KAOP) via a generated wrapper. The imported agent continues running in its original environment while KAOP registers it, captures evidence, and applies governance without code changes.","date":"2026-11-09","dateIsEstimated":true,"signalType":"feature_update","signalTypeLabel":"Feature","sourceUrl":"https://docs.komodor.com/manage-your-agents/build/import-an-agent","publisher":"docs.komodor.com"},{"title":"Komodor Survey Finds More Than 50% of Engineering Organizations Face Control and Governance Issues as They Scale Agentic Operations","summary":"Komodor’s 2026 Agentic Operations Survey found 53.6% of engineering teams cite guardrails as their top challenge when scaling AI agents, with 50.4% concerned about governance and security. Nearly 84% run 10+ agents, yet 75.7% restrict autonomy to read-only or noncritical tasks.","date":"2026-11-05","dateIsEstimated":false,"signalType":"content_marketing","signalTypeLabel":"Content","sourceUrl":"https://komodor.com/blog/komodor-survey-ai-agent-governance-agentic-operations/","publisher":"komodor.com"},{"title":"Komodor in Gartner Research: 2026 Coverage","summary":"Komodor was named in 12 Gartner research documents in 2026, including seven Hype Cycles and the Market Guide for AI Site Reliability Engineering Tooling, positioning it as a representative vendor in AI-augmented SRE and Kubernetes operations. Coverage spans AI for I&O, platform engineering, and cost optimization.","date":"2026-11-05","dateIsEstimated":false,"signalType":"content_marketing","signalTypeLabel":"Content","sourceUrl":"https://komodor.com/blog/komodor-gartner-research-2026/","publisher":"komodor.com"},{"title":"2026 Agentic Operations Survey: How 1,000 Engineering Organizations Run AI Agents in Production","summary":"Komodor published a 2026 survey report on how 1,000 engineering organizations run AI agents in production, highlighting adoption trends, operational challenges, and the need for centralized controls. The report emphasizes that while agent adoption is growing, teams lack visibility, cost control, and governance to safely deploy autonomous agents.","date":"2026-11-05","dateIsEstimated":true,"signalType":"content_marketing","signalTypeLabel":"Content","sourceUrl":"https://komodor.com/resources/komodor-2026-agentic-operations-survey-report/","publisher":"komodor.com"},{"title":"Hillclimbing Meets Production: What Changes When the Agent Runs Your Infrastructure","summary":"Komodor details its Evaluation Factory, a system for continuously improving production agents by measuring outcomes, paths, and costs while adapting to live infrastructure changes. The approach contrasts with Anthropic’s hillclimbing method, emphasizing multi-person oversight, synthetic testing, and living eval sets.","date":"2026-11-04","dateIsEstimated":false,"signalType":null,"signalTypeLabel":null,"sourceUrl":"https://komodor.com/blog/hillclimbing-production-agents-evaluation-factory/","publisher":"komodor.com"},{"title":"The Evaluation Factory: How We Make Agents Measurably Better, Run After Run","summary":"Komodor introduced the Evaluation Factory, a looped system to measure, test, and improve AI agents by comparing versions, running shadow tests, and using LLM judges and golden scenarios. The framework aims to balance precision and cost in production environments while automating insights generation and human decision gates.","date":"2026-10-29","dateIsEstimated":false,"signalType":"feature_update","signalTypeLabel":"Feature","sourceUrl":"https://komodor.com/blog/evaluation-factory-ai-agents/","publisher":"komodor.com"},{"title":"Self-Improving Agents: How to Turn Failures into Fixes Without Babysitting Prompts","summary":"Komodor’s AI team introduced a self-improving agent framework using a teacher-student-evaluator-coach loop to optimize cheaper AI models without manual engineering. In a live demo, a Claude Haiku agent improved from 0.42 to 0.79 accuracy against a 0.85 teacher in three rounds, with only prompt adjustments.","date":"2026-10-28","dateIsEstimated":false,"signalType":"content_marketing","signalTypeLabel":"Content","sourceUrl":"https://komodor.com/resources/self-improving-agents-how-to-turn-failures-into-fixes-without-babysitting-prompts/","publisher":"komodor.com"},{"title":"From Laptop to Production: A Technical Teardown of the Komodor Agentic Operations Platform","summary":"Komodor’s Agentic Operations Platform introduces a two-layer architecture—workers for execution and a control plane for governance—replacing the 'identity, attention, and judgment' lost when agents move from laptops to production. The platform adds guardrails, deterministic workflows, and a Swiss-cheese evaluation model to ensure trustworthy, cost-controlled agent operations across environments.","date":"2026-10-27","dateIsEstimated":false,"signalType":"feature_update","signalTypeLabel":"Feature","sourceUrl":"https://komodor.com/blog/technical-teardown-komodor-agentic-operations-platform/","publisher":"komodor.com"}],"attribution":{"source":"Spyingbee","url":"https://spyingbee.com/updates/komodor","citation":"Spyingbee, \"Komodor updates\", https://spyingbee.com/updates/komodor (retrieved 2026-11-22)"}}