OpenClaw experiments
Local agents, project context and workflow experiments.
Project context
The local agent experiments explore retaining useful project context and reducing repeated steps. Tasks, source material and results remain available for review.
This is development work alongside the website and Motherlabs, rather than a published studio service.
Codex and OpenClaw
Codex is used for software development. OpenClaw is used for local agent and workflow experiments. The projects do not require both tools for every task.
Tool comparisons depend on the specific job, current documentation, supported integrations and maintenance. No general performance comparison is published here.
Experiment scope
An experiment record separates the input, execution and observed result. The local work described here is experimental.
A completed test establishes that run. Continuous operation and reliable customer handling have not been demonstrated by these local experiments.
Mobile requests — proposed
Mobile task requests are a possible extension of the local workflow. A complete route would need to receive the request, run the intended task and return a reviewable result.
An always-on mobile control service has not been verified.
Data paths
A local runtime may still call external models, connectors, email providers or remote tools. Each connection has its own permissions and data handling.
These experiments have not been audited as a local-only system for client information. Client use would require a defined provider, outbound data and retention arrangement.
Permissions
Reading information, drafting a response, sending it and changing an appointment are separate permissions.
The proposed studio use is assistance reviewed by a person. No automatic reply or appointment-control workflow is demonstrated here.