Active development
01 / Infrastructure
Anvil
A foundation your AI can run on.
Automate the path from supported hardware to a private Kubernetes environment, with cluster lifecycle management as a core product responsibility.
Explore AnvilPrivate AI infrastructure & workflows
Put AI to work.
Keep control.
From the first cluster to the first useful agent. Ataraxia is building a private AI platform for the IT team you already have.
Anvil brings order to infrastructure. Atelier is being designed to bring your workflows, policies, and operational feedback into one lifecycle.
Anvil: active development · Atelier: product design
AtelierAgent & workflow lifecycle
BuildGovernRunObserve
Deploy workflows ↓↑ Operational feedback
AnvilPrivate AI infrastructure
ProvisionConfigureMaintain
Compute & GPUs Data & storage
Customer-controlled environment
Optional external model access is a separate, explicit policy decision.
Customer-controlled infrastructure Explicit governance Operational ownership
The problem we’re working on
An AI pilot is only the beginning.
The hard work continues after the demo: connecting data, managing GPUs, defining permissions, handling upgrades, and knowing when a workflow needs attention. Your IT team should have a coherent way to manage that work.
In development · design stage
02 / Workflows
Atelier
A workshop for useful, governed AI.
Design agents around real work. The planned experience connects guided creation, approved tools, human checkpoints, deployment, and observation.
Explore the Atelier conceptOne connected lifecycle
Build. Run. Learn. Improve.
Our goal is to connect infrastructure operations with workflow development, so a problem in production has a clear path back to a reviewed improvement.
Prepare
Provision a supported environment with Anvil.
Create
Define the workflow, sources, tools, and success criteria.
Control
Set permissions and review before deployment.
Improve
Use execution evidence to test and approve the next version.
Target lifecycle. Availability will follow validated product milestones.
Start with real work
A focused first use case.
Begin with a bounded workflow, a named owner, and an outcome you can measure.
Internal knowledge
Find relevant information in approved documentation and return source-backed answers.
View workflow example →Support triage
Organize incoming tickets and prepare responses for an engineer to review.
View workflow example →Incident investigation
Bring operational evidence together to help an engineer assess the next step.
View workflow example →Build with us
Start with one useful workflow.
Bring your infrastructure, your operational constraints, and a problem worth solving. Together, we can define a focused pilot and what success should look like.