Hi there,

Welcome back to Beyond the Prompt!

After the summer break, we have lots of interesting articles for you. Let’s dig in.

💡 From the case studies

We published a fascinating case study about training—yes, training! With AI evolving so quickly, building a growing internal knowledge practice has become really important. This means not only technical skills, but also learning how to solve problems creatively and stay flexible.

If you think your organization could benefit from some guidance in this area, we'd love to hear from you!

📜 From the blog

Caroline shared her approach to tackling AI projects.

I, for one, argued you might want to avoid using OpenAI's agentic workflow builder. You can't say I didn't warn you 😉.

🎙️ From the podcast

It’s Yoann’s turn to talk about his atypical journey from socioligy to AI (🇫🇷 podcast in French)

🧪 From the lab: What we're cooking

One of the most important lab project in the past quarter was reorganizing the lab itself. Our R&D now integrates with knowledge pillars that are identified based on market needs and meaningful technological trends. As a "boutique" player, we must be deliberate in the choices we make, both in pursuing R&D topics, and those we choose to avoid.

🌐 From everywhere else

If you haven't seen this one already, Deloitte has been caught delivering a report riddled with AI-generated hallucinations. A good opportunity that those cases are not the fault of AI, but rather the fault of poor human judgement... like this Californian lawyer.

"Either Embrace AI or Get Out of This Career" is the message adressed developers sent by GitHub CEO Thomas Dohmke.

Will The Fear Of Being Confused For AI Mean That We Will Now Write Differently? I know that before AI, I loved using em-dashes. Not anymore, since it's making it look AI-generated.

Rejoice: with agentic browsers, scammers have a new way to get to your money.

Oh, and hackers can count on "vibe hacking" too!

Delta Air Lines will try to extract more profit with AI-driven pricing.

Anthropic (now valued at $183 Billion - with a B) is entering the financial services game. It's a "contact us" kind of pricing, and I imagine it's intended as a value proposition that targets the €30k-a-year Bloomberg terminals. That's one way to find the funds for a $1,5 Billion book copyright settlement.

I really liked this essay: "AI is a Floor Raiser, not a Ceiling Raiser". It definitely resonates with my personal experience.

This one made me chuckle. All it took for Taco Bell to rethink its AI ordering technology was a dude ordering 18.000 water cups.

While U.S. burns money in search of an elusive "general intelligence", China is taking a more pragmatic approach, seeking efficiencies in agriculture, public services dispatching or education.

OpenAI published an interesting study on consumer usage of ChatGPT. Worth a look if you're interested in the ways it's being used.

👋 Parting thoughts

We just closed our fiscal year. It’s probably meaningless to everyone else, but to us, it’s the end of another chapter in the Agilytic story. We would love for a geopolitically and economically quiet year at some point… but considering the rollercoaster it has been, we feel happy and motivated by what’s ahead.

I hope you are too.

All the best,

Julien

_

PS: Feedback is a gift. Tell me what you liked (and liked less) simply by replying to this email.