AI LAYER
AI steps designed to be bounded, inspectable and reviewable.
The AI layer runs classification, extraction, summarisation and decision-support steps inside workflows. These pages describe product design and demonstrate it locally; no AI service is called.
Examples
AI workflow examples you can try.
AI step types
Assign a label and a confidence score to text, then route on the result.
Input
"My invoice shows a duplicate charge."
These are product capabilities shown as examples. Outputs are static samples, not results from an operating AI service.
Example workflow: support message routing
Simulated in your browserEdit the text or pick a sample. Rules run locally; no AI service is called.
- Customer message
- AI classification
- Priority
- Routing
- Action
Select RUN CLASSIFICATION to send the message through the workflow.
Design
How the AI layer is intended to work.
Provider abstraction
Workflows describe the step, not the model. The provider behind it can change.
Design intent
Structured output
Each step returns data that matches a schema the next step can rely on.
Design intent
Confidence routing
Results below a threshold go to an approval step for human review.
Design intent
Visible inputs and outputs
Run logs show what each AI step received and returned.
Design intent
You stay in control
AI steps assist within a workflow. Rules and people make the final decisions.
Data handling
Controls for what data reaches an AI step will be documented before release.
Planned
In a workflow
Invoice Approval with an extraction step.
The AI step reads the invoice, then rules and an approval decide what happens next.
Invoice Approval
v1 · draftNo executions yet. Select RUN DEMO to simulate a run.
BUILD YOUR FIRST WORKFLOW.
Turn repetitive business processes into reliable, observable workflows.