Getting started with samples¶
WIRE includes sample use cases that you can deploy in one call. Each sample creates a schema inside the app with realistic source data, input views, output tables, configured agents, and a Streamlit viewer to explore the results.
| Sample | Scenario | Agents | Records |
|---|---|---|---|
claims |
Home warranty claims | Triage, analysis, feedback rollup | 125 claims |
underwriting |
Dealer floor-plan loan underwriting | Triage, analysis | 104 loans |
distribution |
Distribution center delivery timeliness | Triage, analysis | 122 deliveries |
Sample data lives entirely inside the app, so you don't need to grant access to any of your own objects.
Deploy a sample¶
Before you start, make sure you've completed After installing, including the warehouse grant.
-- List the samples and whether each one is deployed
CALL wire_native_app.main.list_samples();
-- Deploy a sample, naming the warehouse its viewer runs on
CALL wire_native_app.main.deploy_sample('claims', 'WIRE_XS_WH');
The deploy creates a sample_claims schema in the app and registers the sample's use case
and agents.
Run the sample agents¶
Run triage first, then analysis, which works from the triage results:
CALL wire_native_app.main.run_work_item_agent(
use_case => 'claims', agent_name => 'claim_triage'
);
CALL wire_native_app.main.run_work_item_agent(
use_case => 'claims', agent_name => 'claim_analysis'
);
Each call returns a summary of the run, including how many rows it processed and any errors.
Explore the results¶
Open the sample viewer from the app's page in Snowsight. It shows each work item with its
triage category, confidence, and reason, the analysis results, and a Metrics page with
token usage and, if you granted USAGE_VIEWER, credits and cost.
You can also query the output tables directly:
SELECT id,
results:category::STRING AS category,
results:confidence::STRING AS confidence,
results:reason::STRING AS reason
FROM wire_native_app.sample_claims.triage_output;
Remove a sample¶
Next steps¶
Use the sample as a template for your own use case: see Connecting your data.