Entasis — the subtle convex curvature applied to a column: the correction that makes it appear perfectly straight. Precision that produces apparent perfection.
We build the precision underneath.
Tech lost something.
Not to AI, before AI. The internet used to be a place where people showed up as themselves. Tools had personality. Problems required creativity. Participation meant something because it cost something.
We traded that for ease and speed. Everything got smoother and blander at the same time. Even the tools we use to build things stopped feeling like ours.
Entasis exists to build differently. Not faster. Not easier. More meaningful.
We create tools that augment how you think instead of replacing your thinking. Infrastructure that makes data legible and actionable, not outsourced. Software that has a point of view, because the people who built it had one.
We are not an AI lab. We are a modern data integration and reasoning lab. AI is a lens. Data is the substance. Humans are still the point.
Real people want AI to power them, not replace them. We build the exoskeleton, not the autopilot.
We believe computation — whether over code, knowledge, or thought — can be structured, observable, and interpretable. The machinery should be visible.
Six layers. One coherent system. Each stands on its own — and becomes something more when composed.
-
Medousa — Do the work. A local-first AI workspace on your machines. Tell her what's weighing on you, wake up to finished guides in your vault — not chat that disappears. → medousa.app
-
Resonantia — Navigate the mind. A local-first app for navigating your AI sessions as geometry. Explore the topology of your own thinking, rendered as a constellation across time. Not a summary — a return. → resonantia.me
- Stasis — Long-running AI, finally. A Rust framework for AI systems that behave like distributed applications: durable execution, queue ownership, scheduling, retries, and runtime observability. → runstasis.io
- Grapheme — A governed workflow engine. Language and runtime for policy-scoped automations with explicit control flow, typed transitions, and capability-aware execution. → github.com/EntasisLabs/grapheme
- Locus — The architectural realization of a memory palace. Stitches AVEC, STTP, and retrieval into persistent memory composites. Build extensive memory layouts, the way you envision them. → locus.black
-
STTP (Spatio-Temporal Transfer Protocol) — the typed IR language for cognition. Enables A2A quasi-deterministic communication and verification via structure.
-
ACC (Adaptive Codec Context) — models codebases as living systems, quantifying stability, logic, friction, and autonomy, and providing actionable insight directly in your editor.
-
DMC (Design Meaning Context) — STTP's little cousin for running group studies across different AVEC profiles. STTP keeps what remains true; DMC emerges the meaning behind what was said.
-
instrumenta — protocol-first open tooling for persistent AI memory (STTP) and adaptive code intelligence (ACC). The machinery, not vague magic. → github.com/EntasisLabs/instrumenta
- AVEC (Attractor Vector Encoding Configuration) — the dimensional coordinate system everything else speaks. An OCEAN-inspired take (the Big 5 of psychology) applied to AI context, mapping reasoning posture across Stability, Logic, Friction, and Autonomy.
Together, they demonstrate a new paradigm: AI is not an omniscient oracle. It is a stateless machine operating on biases and structure provided by its creators, computing knowledge into a digital format humans can observe, interact with, and improve.
Herald. Memory. Mind. Understanding.
Every instrument here was built the same way. Out of frustration. Out of needing something that didn't exist yet. From a signal worth chasing.
entasislabs.com · © 2026 Entasis · Apache-2.0