Public-safe prompts
Use generic scenarios and content you are permitted to share publicly.
Visible input guidance and capsSwaves AI Labs · Public-safe experiments
Explore interaction patterns without mistaking a lab for a production service.
Labs is a bounded public environment for demonstrating workflow shapes, interface ideas, and safety behavior with synthetic or public-safe inputs. It does not connect to customer tenants, private repositories, or internal enterprise data.
- Data
- Synthetic, user-entered public-safe text, and approved public material
- Purpose
- Demonstrate interaction and control patterns
- Not for
- Confidential work, consequential decisions, or operational reliance
02
01 · Scope
Labs demonstrates a pattern; it does not certify a production outcome.
Experiments are intentionally constrained so visitors can inspect an interaction without supplying privileged context. Results are illustrative and should be independently verified.
No tenant context
Experiments do not retrieve from customer environments or authenticated enterprise sources.
Separated public interaction pathNo operational action
The lab does not execute changes in customer systems or act as a system of record.
Interaction-only workflow boundary03
02 · Experiments
The useful artifact is the workflow pattern and the questions it exposes.
Current experiments illustrate how a system can gather context, show intermediate stages, present sources, communicate limits, and route a user toward human review.
Ask
Start from a bounded question or representative task without customer-specific details.
Prompt pattern and declared scopeInspect
Observe stages, source cues, and guardrail messages rather than relying on the final text alone.
Visible workflow and source indicatorsReflect
Use the result to identify requirements, failure cases, and production evidence needs.
Questions for a production brief04
03 · Learning
A lab observation is a hypothesis input, not a benchmark result.
Public demonstrations do not represent a customer corpus, identity model, operating load, or review population. Production evaluation requires representative data and accountable reviewers.
Interaction finding
Record whether the interface made evidence, uncertainty, and next action understandable.
Failure question
Identify what could go wrong once private data, permissions, or operational actions are introduced.
Evidence need
State which representative cases and owners would be required for a real readiness decision.
05
04 · Experience
A public demo should reveal its constraints instead of performing certainty.
The experiment surface keeps mode, processing stage, source cues, and safety language close to the generated result so visitors can evaluate the interaction itself.
Illustrative lab surface: scope and limits stay visible beside the interaction.
Mode is explicit
The selected representative workflow remains visible throughout the interaction.
Progress is legible
Stages communicate what the demonstration is simulating or attempting.
Limits stay present
Public-safe guidance and production distinctions are not hidden after launch.
06
05 · Production boundary
A production path begins with customer context that the public lab deliberately excludes.
Enterprise work separately addresses source authority, identity, security, evaluation data, integrations, service ownership, incident response, and change control.
Reframe
Define the actual workflow, consequence, users, decisions, and prohibited behavior.
Re-evaluate
Build representative cases and thresholds with accountable domain reviewers.
Re-engineer
Design enterprise identity, data, controls, operations, and release evidence for the context.
A lab result is a learning artifact, not a production recommendation. Production work starts with a separate risk, data, evaluation, and ownership review.
Open the experiments