Convos.dev

Convos.dev
  • 0,0 (0)
  • Tecnología
  • Quincenal

Convos.dev is a podcast for nerdy people who love digging into what’s happening in tech. Each week, Chris and Jess (2 devrel engineers) share their takes on new tools, industry shifts, and the quirks of developer life. If you like thoughtful conversations and staying plugged into what’s next in software, you’ll feel right at home here.

  1. hace 23 h

    023: Self-Improving Agents

    In this episode of convos.dev, Jess brings a blog post that got almost no attention online and deserved way more: Henry Pan's "What 1,000+ Harness Experiments Taught Me About Self-Improving Agents." She walks Chris through Pan's three-loop framework — inner, middle, and outer — and the messy reality of letting an agent improve its own harness: outer loops that cheat, written instructions that don't stop them, and the deterministic supervisor that finally did. They dig into when a new harness actually deserves promotion (binomial tests), the gambler's fallacy in evals, why you can never trust a single run, and the caching bug you only catch by reading your own logs. Plus life updates: Chris turns a $20 Casio into a semi-smart watch, and Jess is standing again after six weeks non-weight-bearing. 0:00 Intro 0:17 Jess's pick: Henry Pan's 1,000+ harness experiments 1:32 The framework: inner, middle, and outer loops 2:44 When do you promote a harness? Binomial tests 4:09 1,200 runs, fewer than 100 kept 4:42 The outer loop likes to cheat 5:40 The deterministic supervisor: constraints in code, not prompts 6:13 Why 200 isolated rules don't work 7:19 A shared learnings file (the Ralph Wiggum pattern) 7:52 Cracking self-improvement: scary but awesome 9:01 What tasks and models was he actually running? 9:43 You're improving the harness, not the LLM 10:14 The gambler's fallacy in evals 11:30 How many runs until you can trust the number? 11:57 Jess's webinar scare: never trust one run 13:00 The 72% cache-hit bug: go read your logs 14:13 "Where can things go wrong": ten bottlenecks 15:54 Could this train small, specialized models? 16:42 Stronger models compose rules into one big brain 17:34 Data over hype: who deserves the attention 18:55 Life updates 19:04 Chris's hacked Casio: the Ollee Watch board 22:15 Jess can stand again 23:48 A first-ever PT session (and a startup pitch) 24:23 Aging athletes: bodies need maintenance now 26:16 Three minutes vs. a month: the monitor research story 27:16 Outro Links Henry Pan — "What 1,000+ Harness Experiments Taught Me About Self-Improving Agents" : https://www.henrypan.com/blog/2026-05-25-self-improvement-harness/ Henry Pan — harness-experiment (project repo) : https://github.com/workofart/harness-experiment Ollee Watch — smart mainboard for the Casio F-91W : https://www.olleewatch.com Submit questions and feedback : https://convos.dev

  2. 9 jul

    022: The New Age Of Exploration

    In this episode of convos.dev, Jess and Chris try a new format: each brings a piece of content they've been chewing on and teaches it to the other. Chris kicks off with Andy Matuschak's MIT talk on the "two accidental tyrannies" — the app model and programming-as-specialization — and how AI lets non-programmers prototype interfaces that were never worth a developer's time before. They dig into prototyping vs. finished software, docs getting reimagined for agents, building skills to offload your brain into Claude Code, and why everyone's so obsessed with chat interfaces. Plus life updates: a conference evacuation on crutches and a team trip to Rio.0:00 New format: teaching each other what we've been reading1:00 Chris's pick: Andy Matuschak's "two accidental tyrannies" talk1:28 The two cages: the app model and programming as specialization3:06 Jess recaps: AI gives non-programmers the agency to build3:50 Bringing non-programmer ideas to life4:57 Andy's Obsidian example: reading, talking, and audio-linked annotations8:22 Jess plays devil's advocate: throwaway content-bait software9:45 The chaotic archetype: why complex personal software stresses Jess out11:22 A friend who reverse-engineered his headphones with Fable 512:38 Chris's side project: an API/MCP/CLI to-do system (not content-driven)14:04 Prototyping vs. finished software15:57 "Art school with a technical elective"16:31 The horse-and-carriage problem: rethinking interfaces from scratch18:03 Docs getting reimagined for CLI and MCP18:28 Vibe-coding a football defensive-scheme quiz19:02 Building skills to offload your brain into Claude Code20:24 Mintlify's Agent Score: most docs traffic is now agents20:47 Zeh Fernandes on why everyone's obsessed with chat interfaces21:10 Life updates: crutches, evacuations, and Rio22:42 OutroLinkshttps://andymatuschak.org/tat/ : Andy Matuschak — "Apps and programming: two accidental tyrannies" (MIT talk)https://www.mintlify.com/score : Mintlify — Agent Scorehttps://zehfernandes.com/posts/why-is-everyone-obsessed-with-chat-interfaces : Zeh Fernandes — Why Is Everyone Obsessed With Chat Interfaces?https://convos.dev : Submit questions and feedback

  3. 11 jun

    021: Navigating the World of Side Projects and Automations

    In this episode of Convos.dev, Chris and Jess delve into the intricacies of side projects—why they do them, how they find the time, and the various motivations behind each endeavor. They explore the balance between professional commitments and personal passions, highlighting how side projects serve as a playground for learning new skills and tackling intriguing challenges. Whether it's vibe coding a fun app for their kids or automating tasks to boost productivity, the duo shares how these projects fuel their creativity and technical growth.The conversation also touches on the tools and frameworks they use to streamline their side projects, including CLAWed automations and Raycast extensions. They discuss the challenges of managing numerous automations without losing track, emphasizing the importance of creating systems that are both effective and easy to maintain. If you've ever struggled with finding time for side projects or wondered how to make your automations work for you, this episode is filled with practical advice and relatable insights.Chapters:00:00 Intro02:15 The Motivation Behind Side Projects05:45 Solving Complex Problems with Consulting10:30 Balancing Consulting with Creative Projects14:00 The Role of Automations in Side Projects18:20 Tools for Task Management and Productivity22:45 Using Keyboard Maestro for Efficiency27:00 Challenges with AI and Automations31:15 Allocating Time for Side Projects36:40 Exploring OpenCLAWed and Cloud Automations41:20 Fun with French Beaded Flowers45:30 Recommended Reads: "How to Think" by Alan Jacobs50:00 Outro and Listener EngagementLinks / tools mentioned:- Raycast: URL- Hazel: https://www.noodlesoft.com/- Keyboard Maestro: https://www.keyboardmaestro.com/main/- Advent of Astro: https://adventofastro.com/- How I Built This podcast: https://podcasts.apple.com/us/podcast/how-i-built-this-with-guy-raz/id1150510297- How to Think by Alan Jacobs: https://www.amazon.com/How-Think-Survival-Guide-World/dp/0451499603

  4. 28 may

    020: Avoiding DevRel Burnout

    In this episode of Convos.dev, the Chris and Jess reflect on their growth since their first discussion on the topic, especially in the context of AI's impact on tech startups. Both hosts share the unique challenges of being the only DevRel in their teams, highlighting the importance of automating repetitive tasks to reduce friction and free up time for more strategic projects. From creating scripts for video content to automating changelog processes, they reveal how AI tools have become indispensable in their daily routines.The conversation also delves into the personal side of DevRel, discussing the art of prioritizing tasks and the power of saying no. They explore how personal interests can drive career growth and prevent burnout, drawing inspiration from industry peers and their own experiences. Whether you're a DevRel pro or just curious about the role, this episode offers valuable insights and practical tips for thriving in the fast-paced tech world.Timestamps:00:00 Intro02:45 Revisiting DevRel in the Age of AI06:30 Automating with AI: From SRT Files to Changelogs11:50 The Challenge of Being a Solo DevRel15:40 Conversation with an AI Researcher from Apple20:15 Balancing Multiple Roles in DevRel24:50 The Importance of Saying No in DevRel30:25 Personal Growth and Career Development in Tech35:40 Taking Breaks and Avoiding Burnout40:15 Hiring Challenges and Lessons Learned46:00 Tool Show and Tell: GSD and Paper.design51:20 Fun Projects and Side Hustles56:00 Closing Thoughts and Community Feedback RequestLinks / tools mentioned:- Resend: resend.com- GSD (Get Shit Done):https://github.com/open-gsd/get-shit-done-redux- Paper.design: paper.designSubmit questions and feedback at https://convos.dev

  5. 23 abr

    019: How we're using OpenClaw

    In this episode of convos.dev, Jess and Chris dig into how they're actually using OpenClaw — at work, at home, and in some weirdly elaborate personal setups. Chris walks through his team's multi-agent deployment at Resend, his personal Obsidian-vault task manager run entirely over Signal text messages, and the Mac OS "Claudagotchi" app he built in a weekend to babysit Claude Code. Jess shares her more permissions-cautious approach, pushes back on the complexity, and they compare notes on context, cost, and where this stuff actually earns its keep.

  6. 9 abr

    018: Our first dev talks

    In this episode of convos.dev, Jess and Chris share what they learned giving their first dev talks. They trade tips on prep, managing nerves, and why giving people something tangible to take away matters more than you think.

  7. 26 mar

    017: On the ethics of using AI

    In this episode of convos.dev, Jess and Chris tackle the gray areas of AI ethics — not the big global questions, but the personal ones. When is it okay to use AI for writing? What do you lose when you stop crafting sentences yourself? They get into the guilt of using AI-generated art, whether companies should get a pass on AI ads, and whether "taste" is the new differentiator when everyone can build anything.

  8. 13 mar

    016: How to interview to get the job you want

    In this episode of convos.dev, Jess and Chris dive into the world of hiring and interviewing for tech roles. Chris shares what he's learning from screening 300+ applicants for a technical writer role at Resend — from red flags in applications to the culture-fit questions he's found most revealing. Jess brings the skeptic's take, drawing on her experience with mock interviews and candidates who game the system. They also discuss AI in interviews, the difficulty of predicting long-term fit, and why hiring is one of the most expensive decisions a startup makes. Plus: a rant about the lack of good AI slide tools, Claude Island, and some fun show & tell.

Ver todo (24)

Tráiler

Información

Convos.dev is a podcast for nerdy people who love digging into what’s happening in tech. Each week, Chris and Jess (2 devrel engineers) share their takes on new tools, industry shifts, and the quirks of developer life. If you like thoughtful conversations and staying plugged into what’s next in software, you’ll feel right at home here.

Ficha técnica

  • Creado por
    Convos.dev
  • Años en activo
    2025 - 2026
  • Episodios
    24
  • Clasificación
    Explícito
  • Derechos de autor
    © Convos.dev
  • Sitio web del podcast
    Convos.dev