The skinny on what's happening in AI - straight from the previous live session:

Highlights

🤖 Albania Appoints World's First AI Minister

Discussed at [00:17:59]

Albania has become the first country to appoint an AI-powered minister. Diella (meaning "sunshine" in Albanian, and apologies to my Albanian viewers if I butchered the pronunciation on stream!) will handle public procurement decisions, evaluating tenders and managing government spending. Prime Minister Edi Rama says the goal is to make public tenders "100% incorruptible" by removing human decision-makers from the process.

Kyle's take: Initially thought this was marketing nonsense, but it could be brilliant. Using AI for procurement is exactly where you'd start - an AI can't be slipped an envelope under the table. Albania has well-documented corruption issues (repeatedly flagged by the EU), and this could genuinely help.

The AI already powers their e-Albania platform for government services. If they make the decision-making process transparent, this could be a new model for open government. Though obviously there's still room for co

rruption from whoever builds and controls the AI!

Source: Politico

🧠 Claude Gets Memory (But You Probably Can't Have It!!)

Discussed at [00:02:13]

Anthropic is finally rolling out persistent memory to Claude - but only for Teams and Enterprise users. The AI will remember your projects, preferences, and past conversations across chats. They're also adding "incognito mode" for when you don't want something remembered (available to all users now).

Kyle's take: Persistent memory is the killer feature that keeps me using ChatGPT despite preferring Claude's outputs. Without persistent memory, you're re-explaining context every single time - it's exhausting.

The fact they're only giving it to enterprise users first is classic Anthropic - brilliant tech, less brilliant at consumer rollout. Memory creates lock-in - the more an AI knows about you, the harder it is to switch. That's why every AI company should be racing to add this, but they're dragging their feet. Comeon guys!

Source: Anthropic blog post

🏥 AI Safety Hunger Strike Ends After Medical Emergency

Discussed at [00:26:27]

Michael Trazzi ended his 7-day hunger strike outside Google DeepMind's London office after doctors found dangerously low blood glucose and potentially developing acidosis. He experienced two near-fainting episodes in 48 hours before medical intervention. The protest continues with others still striking at Anthropic and DeepMind offices.

Kyle's take: Respect the effort even if I disagree with the cause. The comments on his announcement were brutal - barely any support, lots of people saying he achieved nothing. The original strike got 1.5 million views; this ending got 80,000. Shows how quickly public attention moves on. Hope his health recovers, but I don't think this will change anything at DeepMind.

the torch has been taken up by Denys (see tweets above) but interest seems to have dwindled to basically zero. Day 8’s tweet has 8,000 views. Again, tweet volume isn’t the only arbiter of awareness but it’s a good proxy.

Source: Michael Trazzi

💡 Agent Hack: Auto-Test Discount Codes

Discussed at [00:08:27]

Here's a cool ChatGPT Agent trick from Nick Dobos: "Find discount codes for [store], then go to the site, add an item to cart, go to checkout and test the discount code." I tested it live trying to find codes for an Nvidia RTX 5090 graphics card. The agent found and tested multiple codes automatically, though only found a hilarious 69p discount on a £2,500 purchase (code: "NICE")… not a useful result in this case BUT a useful method!

The important thing here isn’t that the Agent finds the codes. It’s that it then goes to the store and tests them. It goes through the store, basket and checkout and puts in the codes to see if they are valid or not. Very cool!

This saves massive time even if it doesn't find huge discounts. Instead of manually testing 20 dead codes from sketchy voucher sites, the agent does it all for you.

Really cool usage for agents as I know that many people struggle to think of use cases! The trick is thinking of tasks agents can do that would take you ages but are simple enough for them to complete.

Source: Nick Dobos Twitter tip

Member Question from JW: "I have my first demo today with a business for an agent I Vibe coded - it's a research agent that finds leads and contacts for land buyers. Any tips?"

Kyle's response: Brilliant! Start simple - make sure V1 works perfectly even if it's basic. Too many people try to build the entire CRM on day one and it falls apart. Get them a working prototype, get testimonials, then upsell the £10k version later. Remember: zero to one client is harder than one to a hundred. Also, that first positive feedback is gold for winning other clients. Good luck - let us know how it goes!

This question was discussed at [00:16:43] during the live session..

Want the full unfiltered discussion? Join me tomorrow for the daily AI news live stream where we dig into the stories and you can ask questions directly.

Streaming on YouTube (with screen share) and TikTok (follow and turn on notifications for Live Notification).

Audio Podcast on iTunes and Spotify.

AI with Kyle - Tools, Resources and Products: Daily Show | Build an AI Business Challenge | AI Playbooks | 10 Week AI Summer Camp | BATON AI Business Series | AI App Building | AI Trainer Business Planner | What AI Business Should You Start? | AI Authority Accelerator | AI Automation Accelerator | 🏁 AI Workshop Kit