blog

what we're thinking, building, and learning along the way.

thinkingwenzel orlandwenzel orland2026-04-034 min

expertise is the new asset

as LLMs commoditize, the differentiator is the expertise you feed them. taste, judgment, decision-making. the things that can't be downloaded.

engineeringwenzel orlandwenzel orland2026-03-286 min

why memory changes everything

most AI systems forget everything between conversations. we built a 3-layer memory architecture that doesn't. here's how it works and why it matters.

companywenzel orlandwenzel orland2026-03-214 min

building in europe, on purpose

we're a german GmbH building EU-sovereign AI infrastructure. it's a competitive advantage that US companies can't replicate.

engineeringwenzel orlandwenzel orland2026-03-155 min

the agent quality problem

everyone's building agents. almost nobody's measuring whether they're actually good. evaluation is the difference between a demo and a product.

companywenzel orlandwenzel orland2026-03-083 min

why we open source the core

our core platform is open source. the best infrastructure gets adopted when people can see how it works.

thinkingwenzel orlandwenzel orland2026-02-284 min

AI won't replace experts. it will make them infinite.

the question is whether you can capture expertise before it walks out the door.

industrywenzel orlandwenzel orland2026-02-204 min

the real cost of stateless AI

every time you re-explain context to an AI, you're paying for it. in tokens, in time, and in quality. stateless AI is expensive in ways most people don't measure.

thinkingwenzel orlandwenzel orland2026-02-123 min

what we mean by ownership

you own the agent. you own the memory. you own the outputs. if you leave, everything comes with you. that's the point.