Poesis
Usage / Use case Implemented
Qualify model routing for a workload
Compare candidate model routes against workload-specific quality, cost and disclosure criteria using retained evaluation evidence.
Platform
What serves this use case
Features
- Agentic Harness: LLM resolution Implemented
Values
Values composed with this use case
- Value Implemented
Qualify delivery hosts and model routes
Evaluate host instruction delivery and model routes against explicit workload quality, cost and disclosure criteria rather than assuming interchangeable behavior.
Supported by Capabilities(1)
Supported by Features(2)
- Value Partial
Control sourcing data exposure.
Bound the planned sourcing workflow with explicit policies for tool/model egress, credentials, telemetry and retention, including any data leaving the local process.
- Value Partial
Control local delivery data.
Choose local tools and Git workflows, with explicit policies for model/tool egress, telemetry, remotes and backups.
Supported by Affordances(1)
Supported by Capabilities(2)
Supported by Features(2)
- Value Partial
Keep artifacts in a chosen working tree.
Own the Git working tree and its artifact history, with separate disclosure policies for journals, remotes, tools, models and backups.
Supported by Capabilities(1)
Supported by Features(2)
Pains
Pain points
- Pain point Partial
Host or model selection changes behavior without verifying workload quality, instruction delivery or disclosure limits.
Required guidance and checks disappear or unsuitable model routes produce costly, untrustworthy work.
Addressed by Capabilities(1)
Addressed by Features(3)
Experienced by Actor types(3)