$ ls -la ./daily-media-update/
// experiment: 10 Mar → 13 Mar 2026 · concluded
A 4-day automation experiment that wired together a personal AI morning briefing pipeline from scratch.
// how it worked:
- → GitHub Actions cron fired every morning at 7am
- → GPT-4o read that day's reminders, emails, and AI news headlines
- → Wrote a John Oliver-style monologue from the combined context
- → OpenAI TTS (fable voice) converted it to audio
- → Output committed to the repo · page rebuilt · deployed to Vercel
// result: a fully automated daily briefing — no manual steps, end to end.
$ cat ./archive/
// Friday, March 13, 2026
transcript
Hello there! It’s Friday, March 13, 2026. Yes, Friday the 13th. A day that's essentially a full-scale horror movie set to elevator music. Let’s dive straight into today’s terrifying saga of reminders, shall we? First off, you have the gym at 7am. Yes, the gym. Nothing says "happy Friday" like self-inflicted physical torture. Then it’s time to call the dentist and reschedule because, apparently, dental punctuality is now equivalent to a U.N.-level summit arrangement. If that wasn’t enough adrenaline, you also need to review PR for a couple task manager. Which sounds like an app designed to ensure that you and your significant other bicker efficiently. Also, DSA practice: two problems. Because what’s more exhilarating than algorithm anxiety before breakfast? On to emails from Rahul Rathod. Let's marvel at this modern Fitzgerald of the inbox who has penned emails with the captivating subject lines: 'Gym', 'Flight check', 'dental appointment', and my personal favorite, 'hey there.' It’s less like reading emails and more like unraveling post-modern poetry. Now, for today's AI news that sounds totally reasonable and not at all dystopian. Facebook Marketplace has employed AI auto-replies for those wonderfully tedious 'Is this still available?' messages. So, the only thing more robotic than Facebook itself is now, in fact, Facebook. Over at Gemini, task automation is now a thing. Yes, because the idea of an AI handling your dinner and transport doesn’t scream: 'bring on the futuristic apocalypse'. Finally, Anthropic has gifted us with Claude AI, which can now reply with charts and diagrams. Fabulous. So we've officially reached an age where you'd need an AI to decipher the AI's diagrams. What a time to be alive. And as we wrap up this rollercoaster of a morning, remember to tackle your tasks with the fervor of Friday the 13th itself, and remember that even Friday the 13th once a month means you’re probably still winning. Off you go, my brave little proactivity warriors. You’ve got this, until maybe everything else. See you next time!
// Thursday, March 12, 2026
transcript
Good morning. It's Thursday, March the 12th, 2026. Otherwise known as: welcome to the day that’s almost a Friday but not quite, and therefore has no right to exist. Now, let’s begin with reminders. You have a gym session at 7am — upper body. Because nothing says "I have control over my life" quite like doing bench presses next to someone grunting like a slightly asthmatic rhinoceros. Call your mum back. She’s been waiting. Meanwhile, the solving of binary search problems awaits. And if you thought shifting boxes in a code base was mundane, well, you’ve got a couple of task manager PRs to push. Treat THEM like they’re about to trigger a war with Norway. Emails, then. You’ve deployed your project 'rahul-blog' to production. Exciting? Maybe. Useful? Uncertain. Then LinkedIn tells you that you appeared in 47 searches this week. Recruiters? Or just random bots infinitely more curious about whether you prefer JavaScript over Python? And finally, Notion’s gentle nudge that their trial is ending soon. Because those clever digital organisers are all about binding you in with alluring niceties, only to slap a credit card bill across your face like a wet fish. In AI news, Anthropic just raised $3 billion, valuing the company at $60 billion. That's right, a fortune, for the company preaching AI SAFETY while possibly using the funds to build more AI. It’s the fiscal equivalent of building a fire department using flamethrowers. AI tutors now smash human teachers in test prep, although amusingly, standardised tests may not be relevant metrics. The AI understood this too. It’s like hiring Usain Bolt to win a walking race because his legs technically move the fastest. And get this: Microsoft Copilot is now in Windows Task Manager. So, you can ask AI why your PC crawls like molasses. Spoiler: it’s using all its time reassuring you with scientific truths. Remember, you’ve got this. We’ve all got this. But if, like AI, we've no idea what 'THIS' truly means, then at least we'll be confused together with chiseled arms and possibly the lingering scent of failure masking itself as progress. Have a decent day. Sort of. Please.
// Wednesday, March 11, 2026
transcript
Today is Wednesday, March 11, 2026. Or as we call it here at Last Week Tonight, one day closer to Friday. But for the love of all that is good, can we possibly do something about Tuesday? Now, let's face today’s litany of reminders. Gym: leg day. Because nothing, I repeat, nothing screams 'start of potential injury' quite like throwing yourself into squats before you've had coffee. And do remember to pay that electricity bill. Because every good hero's story begins with a humble glow, not the flicker of impending darkness. Add to that a dash of dynamic programming practice—just in case tackling intricate algorithms while reading a chapter on system design doesn’t sound enough like a thrilling third act. Emails. Right, we’ve got Amazon chiming in with that joyous 'Your order has been dispatched' update. Keyboard stand landing Thursday if you're feeling jazzy. Then there’s the GP's reminder for tomorrow's appointment. Because nothing says 'Wednesday motivation' like medical bureaucracy. Rahul, from Rahul's Bank, delivers the most cheery note of all: the low balance alert. Just a friendly nudge to keep you frugal lest you tip the scales into total financial ruin. In the fantastical world of AI, OpenAI introduces a new feature allowing artificial intelligence to autonomously operate businesses. Which is brilliant news for founders, and prompts an industry-wide scramble for employees to polish those LinkedIn profiles. Google's AI is graciously rewriting your emails to make you sound more 'professional'—a euphemism for turning you part-robot in corporate correspondence. More 'How do you do?' and less 'What's up?' apparently. And, researchers highlight that language models are hallucinating 23% of their cited academic papers. Academia is aghast, while the rest of us, let's be honest, have been hallucinating the credibility of half-baked research for years. So, go forth into this marvelous Wednesday. Polish your CV. Check your bank balance. Maybe learn dynamic programming—but for heaven's sake, don’t skip leg day. Because if we’re going to limp through this week, let’s have a solid reason why.
// Tuesday, March 10, 2026
transcript
It's Tuesday, March 10, 2026. Because life still insists on using actual dates, like we’re all keeping appointment diaries from the 1800s. Now, if you're planning a morning run at 6:30 am, congratulations on being an overachiever. But remember, the true marathon is submitting your tax documents, because if there's one thing more terrifying than running, it's paperwork. In your emails, GitHub has alerted you to a new sign-in from Sheffield. Which is either someone trying to steal your code or it's that peculiar form of cyber-travel where your laptop spontaneously takes a vacation. Meanwhile, Netflix has charged you GBP 10.99. For those keeping track, that’s roughly the GDP of a small country each month, all for the privilege of endless scrolling. And a quick note from Dad about your car's MOT. Because in the world of existential threats, your car failing an inspection is apparently up there with North Korean missile tests. In the world of AI, Mistral AI has released a model that beats GPT-4o on benchmarks. If you’re wondering if benchmarks mean anything these days, don’t worry, the researchers are wondering the exact same thing. Meanwhile, Meta has unveiled AI glasses that recognize people in real-time. They call it a feature; everyone else calls it terrifying. What a dreadful time to forget someone’s name. And perhaps predictably, AI coding tools now write 40% of code at major tech companies. Senior engineers apparently review it. And by review, I assume they mean skim it with about as much focus as they give their coffee orders. There you have it. As you head out into the world today, remember: we're just cogs in a global machine that occasionally glitches, and if you're lucky, it might even run Windows 95. Keep going. You've got this... probably. Good luck.