# sponsor-dynatrace

Published articles for sponsor-dynatrace.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## It passed CI. It passed your evals. The customer still got the wrong answer.

DevFeed: [It passed CI. It passed your evals. The customer still got the wrong answer.](<https://devfeed.tech/articles/it-passed-ci-it-passed-your-evals-the-customer-still-got-the-wrong-answer-10828.md>)

Original publisher: [Read original article](<https://thenewstack.io/ai-agent-trace-debugging/>)

Author: Sean O'Dell

Published: 2026-09-13T14:00:00Z

Content type: article

Language: en

Sources: [The New Stack](<https://devfeed.tech/sources/the-new-stack.md>)

Topics: [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [agent observability](<https://devfeed.tech/topics/agent-observability.md>), [observability](<https://devfeed.tech/topics/observability.md>), [tracing](<https://devfeed.tech/topics/tracing.md>), [dynatrace](<https://devfeed.tech/topics/dynatrace.md>), [ci](<https://devfeed.tech/topics/ci.md>), [coding](<https://devfeed.tech/topics/coding.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-engineering](<https://devfeed.tech/tags/ai-engineering.md>), [ci](<https://devfeed.tech/tags/ci.md>), [coding](<https://devfeed.tech/tags/coding.md>), [dynatrace](<https://devfeed.tech/tags/dynatrace.md>), [observability](<https://devfeed.tech/tags/observability.md>), [post-contributed](<https://devfeed.tech/tags/post-contributed.md>), [sponsor-dynatrace](<https://devfeed.tech/tags/sponsor-dynatrace.md>), [sponsored](<https://devfeed.tech/tags/sponsored.md>), [sponsored-post-contributed](<https://devfeed.tech/tags/sponsored-post-contributed.md>), [traces](<https://devfeed.tech/tags/traces.md>)

### AI overview

The article explains how AI-agent failures can pass CI and evaluations while still producing slow or incorrect customer-facing results. It presents distributed traces and agent trajectories--model calls, tool calls, arguments, and results--as evidence for debugging retrieval behavior, release context, and feature-flag state.

### Source excerpt

A diff is not evidence. It's a statement of intent. The tests passed. The review's done. The change is live. The post It passed CI. It passed your evals. The customer still got the wrong answer. appeared first on The New Stack.

## The AI-native SDLC won't be one process

DevFeed: [The AI-native SDLC won't be one process](<https://devfeed.tech/articles/the-ai-native-sdlc-won-t-be-one-process-8862.md>)

Original publisher: [Read original article](<https://thenewstack.io/spec-driven-sdlc-gates/>)

Author: Anirudh Ramanathan

Published: 2026-09-12T14:00:00Z

Content type: opinion

Language: en

Sources: [The New Stack](<https://devfeed.tech/sources/the-new-stack.md>)

Topics: [AI Bots](<https://devfeed.tech/topics/ai-bots.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-engineering](<https://devfeed.tech/tags/ai-engineering.md>), [anthropic](<https://devfeed.tech/tags/anthropic.md>), [compliance](<https://devfeed.tech/tags/compliance.md>), [devops](<https://devfeed.tech/tags/devops.md>), [dynatrace](<https://devfeed.tech/tags/dynatrace.md>), [github](<https://devfeed.tech/tags/github.md>), [kiro](<https://devfeed.tech/tags/kiro.md>), [platform-engineering](<https://devfeed.tech/tags/platform-engineering.md>), [post-contributed](<https://devfeed.tech/tags/post-contributed.md>), [sdlc](<https://devfeed.tech/tags/sdlc.md>), [signadot](<https://devfeed.tech/tags/signadot.md>), [spec-driven-development](<https://devfeed.tech/tags/spec-driven-development.md>), [sponsor-dynatrace](<https://devfeed.tech/tags/sponsor-dynatrace.md>), [sponsor-signadot](<https://devfeed.tech/tags/sponsor-signadot.md>), [sponsored](<https://devfeed.tech/tags/sponsored.md>), [sponsored-post-contributed](<https://devfeed.tech/tags/sponsored-post-contributed.md>)

### AI overview

The article argues that AI-native software development needs risk- and accountability-based process variants rather than one fixed, spec-driven workflow. It emphasizes deterministic policy enforcement, agent self-checks, human approvals, and auditable records.

### Source excerpt

Anthropic recently published its AI-Native SDLC Playbook. Its central claim is that "code is no longer the bottleneck." When agents The post The AI-native SDLC won't be one process appeared first on The New Stack.