About MyFirst.News
Hi, I’m Max, a dad and software developer. I started MyFirst.News after my eight-year-old told me, “Daddy, I want to listen to news.”
This is the actual news, explained so curious kids can understand what happened, why it matters, and what is still uncertain—without talking down to them or using graphic detail.
The editorial promise
Episodes focus on impactful world, science, technology, and civic stories. Facts come from official sources and reputable reporting. Developing claims are attributed, uncertainty stays visible, and local crime is generally left out.
Sam and Leah are the two co-hosts. Each day, they pick the most important story and unpack it: what happened, the background, and why it matters. One host explains while the other asks the questions a kid might have. Episodes run about five minutes and end with a question families can discuss together. New episodes are made daily, including weekends.
How it’s made
My local AI agent handles story discovery, research, drafting, transcript review, speech generation, audio quality checks, and publication. Each episode must pass recorded transcript, audio, feed, page, and media checks before it goes live.
Public episode pages include concise show notes and source links.
A note for parents and caregivers
Every child is different. Please use your own judgment about whether an episode is appropriate for your child. You may want to listen first or listen together.
Change history
How episodes have changed since the first one, newest first.
- : The co-host’s questions now follow what a kid would most want to know at each moment, and surprising details get explained as soon as they come up. Listener feedback said some questions skipped ahead or assumed things the episode hadn’t explained.
- : Stories now start from something familiar, often how things work in America, and build up one step at a time before getting to the day’s news. Each episode covers fewer details to make room. Listener feedback said stories were still hard to follow, even with hard words explained.
- : Tried a different AI model, Claude Opus 5.5, for researching and writing each episode, to compare its writing style with GPT-6.1 Sol’s.
- : Switched research and writing back to GPT-6.1 Sol to reduce the cost of making each episode.
- : Tried a newer AI model, Gemini 4 Argon, for researching and writing each episode. Sam and Leah’s voices also moved to Google’s newer voice model, Gemini 3.8 Flash TTS, because Google is retiring the old one in November.
- : Each episode now explains one important story in about five minutes, instead of three stories in eight to ten. One host leads the whole episode while the other asks questions. Based on listener feedback, this goes deeper on the day’s most important story in a shorter episode. I also switched to newer AI models: Grok 4.7 now researches and writes each episode, and GPT-6 Astra reviews each script.
- : Tried a newer AI model, GPT-6.1 Sol, for researching and writing each episode.
- : Tried a newer AI model, GPT-6 Astra, for researching and writing each episode.
- : Listeners asked for a fuller explanation of Russia’s war against Ukraine, so I released the first special episode, a ten-minute explainer on it. Specials are extra releases, so the regular daily episode still comes out that day.
- : After trying GPT-5.6 Terra, moved research and writing back to GPT-5.6 Sol.
- : Replaced the roundup of four to six stories with one main story explained in depth, plus two shorter ones, so the day’s most important story gets the most time and context. Sam and Leah also started taking turns leading the main story from one day to the next.
- : My AI agent started making editorial calls and publishing episodes on its own. Before that, I approved each script and listened to each episode before it went out. The automated checks had proven reliable enough to trust, and dropping the manual pass lets a new episode come out every morning. For most of the next week, I also tried GPT-5.6 Terra for researching and writing episodes.
- : Added a longer pause between stories, after listener feedback that it was hard to tell when one story ended and the next began.
- : First episode: four to six stories in up to ten minutes. One host led each story while the other asked questions, and every episode ended with a question for families to talk about. GPT-5.6 Sol researched and wrote each episode, and Kimi K3 reviewed each script.