{"generated_at": "2026-05-29T22:09:45.155113Z", "slug": "how_i_ai", "source_id": "src_how_i_ai", "name": "How I AI", "episode_count": 3, "avg_signal": 69.2, "median_signal": 73.0, "top_signal": 80.4, "latest_episode_at": "2026-05-28T21:32:37Z", "earliest_episode_at": "2026-05-25T11:00:00Z", "category_mode": "ai", "cover_image_url": "https://d3t3ozftmdmh3i.cloudfront.net/staging/podcast_uploaded_episode/43412682/43412682-1780002704196-ce78506b5ffd7.jpg", "rank_score": 94.801, "episodes": [{"episode_id": "ep_how_i_ai_6242df1851aa", "episode_title": "Claude Opus 4.8 is here. Is it as good as they say?", "podcast_name": "How I AI", "podcast_slug": "how_i_ai", "source_id": "src_how_i_ai", "category": "ai", "publish_date": "2026-05-28T21:32:37Z", "overall_score": 80.4, "score_breakdown": {"clarity": 85.0, "originality": 82.0, "hype_penalty": 2.0, "actionability": 75.0, "technical_depth": 78.0, "information_density": 72.0}, "podcast_cover_url": "https://d3t3ozftmdmh3i.cloudfront.net/staging/podcast_uploaded_episode/43412682/43412682-1780002704196-ce78506b5ffd7.jpg", "source_link": "https://podcasters.spotify.com/pod/show/pen-name/episodes/Claude-Opus-4-8-is-here--Is-it-as-good-as-they-say-e3k1l0v", "audio_url": "https://anchor.fm/s/1035b1568/podcast/play/120689119/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-28%2F425086059-44100-2-ee030741560df.mp3", "listen_url": "https://podcasters.spotify.com/pod/show/pen-name/episodes/Claude-Opus-4-8-is-here--Is-it-as-good-as-they-say-e3k1l0v", "verdict": "worth_your_time", "why_listen": "You\u2019ll learn where Opus 4.8 excels in agentic coding and where its overconfidence can mislead, with real-world test results from prototype to production edge cases.", "summary": "Claude Opus 4.8 shows strong performance in greenfield coding tasks and one-shot feature development, shipping working code in a prototype environment. However, it struggles with edge cases, existing codebases, and long-horizon autonomy, often hallucinating or over-rotating on small data points without validation. Despite high benchmarks like 69.2% on Sweebench Pro, the model exhibits overconfidence ungrounded in data, particularly in strategy and complex refactoring tasks."}, {"episode_id": "ep_how_i_ai_8108bcb9bff7", "episode_title": "The Codex feature that works while you sleep", "podcast_name": "How I AI", "podcast_slug": "how_i_ai", "source_id": "src_how_i_ai", "category": "ai", "publish_date": "2026-05-27T11:00:00Z", "overall_score": 54.2, "score_breakdown": {"clarity": 0.0, "originality": 94.0, "hype_penalty": 0.0, "actionability": 0.0, "technical_depth": 82.0, "information_density": 75.0}, "podcast_cover_url": "https://d3t3ozftmdmh3i.cloudfront.net/staging/podcast_uploaded_episode/43412682/43412682-1779833292030-9dd15d6ecf581.jpg", "source_link": "https://podcasters.spotify.com/pod/show/pen-name/episodes/The-Codex-feature-that-works-while-you-sleep-e3judoo", "audio_url": "https://anchor.fm/s/1035b1568/podcast/play/120583384/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-26%2F424944949-44100-2-5bf9048d90ac.mp3", "listen_url": "https://podcasters.spotify.com/pod/show/pen-name/episodes/The-Codex-feature-that-works-while-you-sleep-e3judoo", "verdict": "must_listen", "why_listen": "Learn how to offload persistent, tedious tasks to AI by defining measurable goals\u2014enabling autonomous problem-solving over hours without supervision.", "summary": "Goals in Codex enable long-running, autonomous AI execution by defining measurable outcomes instead of one-off prompts. The feature works by iteratively testing, executing, and validating steps until a predefined success state is reached\u2014such as eliminating all errors in a codebase or cleaning an inbox. Real-world examples include resolving thousands of edit-operation errors in a document editor and automating email triage across 4,000 messages."}, {"episode_id": "ep_how_i_ai_6328234be250", "episode_title": "How the engineer behind Claude Cowork actually uses Claude | Felix Rieseberg (Anthropic)", "podcast_name": "How I AI", "podcast_slug": "how_i_ai", "source_id": "src_how_i_ai", "category": "ai", "publish_date": "2026-05-25T11:00:00Z", "overall_score": 73.0, "score_breakdown": {"clarity": 75.0, "originality": 85.0, "hype_penalty": 3.0, "actionability": 80.0, "technical_depth": 62.0, "information_density": 58.0}, "podcast_cover_url": "https://d3t3ozftmdmh3i.cloudfront.net/staging/podcast_uploaded_episode/43412682/43412682-1779469328660-24f2ec3ed82cd.jpg", "source_link": "https://podcasters.spotify.com/pod/show/pen-name/episodes/How-the-engineer-behind-Claude-Cowork-actually-uses-Claude--Felix-Rieseberg-Anthropic-e3jn9c2", "audio_url": "https://anchor.fm/s/1035b1568/podcast/play/120349506/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-21%2F424637135-44100-2-74cda3214bf81.mp3", "listen_url": "https://podcasters.spotify.com/pod/show/pen-name/episodes/How-the-engineer-behind-Claude-Cowork-actually-uses-Claude--Felix-Rieseberg-Anthropic-e3jn9c2", "verdict": "must_listen", "why_listen": "Learn how an Anthropic engineer leverages Claude to build maintainable, data-driven personal systems while avoiding the pitfalls of over-automation.", "summary": "Felix Rieseberg, an engineer at Anthropic, uses Claude to solve real-world personal and product problems by treating data as a first-class citizen, building abstractions over time, and integrating AI into workflows without over-automating. He emphasizes comfort with latency, structured data extraction from silos like email, and using AI as a collaborative thought partner rather than a cursor-mover. His approach centers on iterative task abstraction, long-term maintainability, and designing systems where AI augments human judgment."}], "category_breakdown": [{"category": "ai", "count": 3}]}