I outsourced my AI news briefing to AI

I wanted to keep up with AI without spending every morning drowning in AI news. So I built an AI whose job is to read the AI news for me.

There is too much AI news.

This is partly because AI is moving ridiculously quickly, but also because the AI industry has developed the communications discipline of a LinkedIn influencer who's just discovered they can add emojis to their posts.

Everything is a launch. Everything changes everything. Every Tuesday, apparently, the future arrives again. Oh and the human race is doomed to extinction and it's getting closer (see my AI Apocalypse Countdown).

I actually work in AI - so I genuinely do need to keep up - it's just that I don't particularly want to spend the first hour of every morning opening twelve newsletters, three frontier model blogs, LinkedIn, yada yada yada.

So I did the obvious thing. I outsourced keeping up with AI to AI.

My robot reads the internet before I wake up

I’ve been building a personalised AI briefing that does the boring bit for me.

Every morning, the system goes looking for what has changed. It has access to my email inbox, so it checks the AI newsletters I'm subscribed to. It searches the web for developments those newsletters might have missed. It ranks stories according to what I actually care about, strips out duplicates, tries to separate meaningful developments from launch-day froth, and turns the result into one briefing delivered directly into my inbox at 6am.

The important word there is personalised. For me personally, I'm interested in the genuinely exciting news, capability advances & the more curated it is, the better blogs I'm able to write for you folks. For my job, I’m generally interested in keeping up with agentic AI, enterprise AI architecture, AI adoption, AI operating models, task economics / AI FinOps, governance, new model capabilities and what any of this actually means for companies trying to use AI.

I also ask the final part of the update to be an AI lesson for me - and give chatGPT free rein to choose what to teach me in, but it has to be brief, informative & insightful. Just in the last few days I've had master-classes waiting in my inbox on AI FinOps, AI Residency vs Sovereignty, RAG vs fine-tuning, Agentic architectures etc etc.

How is it set up? Pretty basic - it's a scheduled job in chatGPT which I've prompted to make sure it's right. I've even given it word limits as sometimes it tends to get a bit carried away (like me..?). I had a bit of argy-bargy with the email deliverability as Outlook kept thinking it's some form of AI spam.

So is it any good?

In short: absolutely yes. I would recommend this to everyone and it's simple to set up.

Should this replace you reading blogs & newsletters? Well, certainly not all of them - or at least prioritise still reading the good one's (hint hint). Yes AI is good at curating and aggregating information but I treat my 6am AI briefing as a bit more of a 'primer'. It hasn't yet replaced that feeling of reading a decent article/ post and coming away knowing something brand new.

But I do think this points towards a shift in how we consume information. AI can increasingly decide not merely what information exists, but what information I should probably understand next, given what I already know and what I’m trying to become better at. It feels closer to having a research assistant than subscribing to a newsletter.

So have a go - get your 6am AI briefing and see if it's any good. Ensure you system prompt it to prioritise Oliphonic of course...