Thinking in
Systems & Context
In-depth explorations of enterprise context management, AI-powered workflows, and the engineering patterns that make complex software actually work. Written by practitioners, not theorists.
BLORPBLORP Foundations
Foundations is the architecture pattern and reference implementation BLORPBLORP deploys to make enterprise AI projects ground out, scope cleanly, trace end-to-end, and integrate with the rest of your stack. Five pillars, one product.
Grounded retrieval
Most enterprise AI projects fail because the model answers from training data instead of your actual documents. Grounded retrieval is the discipline that fixes that.
Context management
A grounded model still needs to know who is asking, what they are allowed to know, and what they said in the previous turn. Context management is the layer that handles all three.
Multi-model routing
A single AI provider is a single point of failure, a single price ceiling, and a single capability ceiling. Multi-model routing fixes all three.
Provenance + observability
When the AI tells a customer something, you need to be able to show — months later — exactly why. Provenance and observability are how Foundations makes that possible.
Mesh integration
A grounded, traced, scoped AI that lives in a tab nobody opens is still a failed deployment. Mesh integration is what makes the AI a participant in your actual workflows.
The Foundations-First Approach to Enterprise AI
From the road, both houses look identical. Only when the storm rolls in does the difference become visible. Most enterprise AI is built on sticks — context scoping, compliance attenuation, audit trails, and hallucination posture are the four bedrock layers that determine whether your system survives audit, regulator, or breach.