Projectious explores and builds the technical and organizational layers required to turn emerging AI capabilities into reproducible, governed, and auditable ways of working.
The work connects operating-model design—responsibilities, controls, repeatability, sourcing, and auditability—with concrete technical mechanisms. Some repositories are usable tools, others are applied research or explicitly labelled prototypes. Their status and limitations are part of the evidence.
| Project | What it demonstrates | Status |
|---|---|---|
| aibox | Reproducible, terminal-first AI development workspaces generated from a declarative project contract | Usable project — active development |
| processkit | Provider-neutral process memory, validated project state, skills, and MCP tools for coding agents | Usable project — active development, pre-1.0 |
| ai-market-research | Sourced, inspectable decision support for models, harnesses, subscriptions, and self-hosting choices | Applied research — actively maintained |
| kubeclaw | A learning prototype exploring infrastructure isolation and network policy for agent workloads | Working prototype — not production ready |
Additional experiments and supporting assets include kaits, ainfra, the Projectious Brand project, and the public website. Experimental repositories should be read according to the status and limitations stated in their own README.
secure infrastructure targets
│
▼
reproducible workspace images and deployments aibox
│
▼
governed project memory and agent workflows processkit
│
▼
evidence-based technical and sourcing decisions applied research
The projects are deliberately separated by responsibility. Infrastructure templates provision targets; aibox builds and deploys workspaces; processkit owns process content and its installation; research repositories make decision evidence inspectable.
Projectious is led by Bernhard Gerlach, an Operations, Transformation, and Technical Program leader extending operating-model expertise into AI-assisted delivery, agent infrastructure, and reproducible technical workflows.
AI tools assist parts of the research, implementation, and documentation. Problem framing, architecture boundaries, portfolio decisions, review, and the standard of evidence remain human responsibilities. Each repository should document its own validation and limitations rather than relying on this profile for maturity claims.
The presence of a repository does not imply customer adoption, production operation, security assurance, or enterprise support. Prototype and research labels are intentional. Follow each project’s quick start, tests, releases, and limitations for the current evidence.
Learn more at projectious.work.