AI Blueprint | Aurelio AI

Our AI blueprint reference is an evolving document listing our best practices for building AI applications.

Although this combination of technologies has proven to be incredibly powerful and adept across a broad range of use-cases, we need to state that there is no perfect one-size-fits-all. The blueprint is designed to be tweaked and rearranged where required. As such, we focus on modular technologies rather than monolithic monsters. We have worked with many monolithic AI-frameworks in the past and our conclusion every time has been that it is not worth it.

Here is our go-to blueprint for AI projects.

Core Tech

graphai as a foundation for building graph-based AI agents / workflows. We like that GraphAI does not provide any abstractions for LLMs, embeddings, vector DBs, etc. We treat it as a foundation for which we build our own use-case-specific and ultrafast AI-framworks on.

API / DB Layer

FastAPI
Pydantic
Sqlalchemy / SQLModel
Alembic
Postgres
Livekit for advanced streaming and voice capabilities via their RTC infrastructure.

CI/CD

Dagger

Infrastructure

Kubernetes
Terraform
Kustomize
GCP or Hetzner preferred