LinkedIn·Tuesday, 25 August 2026·1d ago

For the first time since the current AI rush started, our team has stopped reaching for the most expensive model by default. While we are…

Jeroen Van Hautte

Co-Founder and CTO at TechWolf 🐺 | AI-first work | Building the context graph for the workforce

For the first time since the current AI rush started, our team has stopped reaching for the most expensive model by default. While we are still enthusiastic users of Fable and Sol, people are now choosing smaller models often too. Sometimes they do it to stretch their budget, sometimes it's just for speed. It's a broader trend too: routers like OpenRouter are growing rapidly based on the same premise, and it's proving that lots of the work can now be done with cheaper models. More is coming too: Ox Alpha, the experimental model on there that has been receiving a ton of buzz recently, is currently being provided for free at a stunning 100T tokens per day. When a new level of intelligence becomes abundant, most of us need to go back to the drawing board. Things that were cost prohibitive just months ago are rapidly becoming affordable, and for the first time that includes genuinely complex problems. For TechWolf, that mostly means a shift from agents assisting to agents acting much more proactively and independently. It means agents can find opportunities and work on them before getting any human approval. It means the relationship people have with their tools and products will find a whole new dynamic. All of this is bringing us back to the same thing again: to ground this exploding level of agent activity, you need a great context layer, or else you can expect chaos. There's a real advantage to being early here: it helps all those agents push in the same direction.

♥ 37💬 2

View on LinkedIn