Docs

How to stand up your first SparkWren workflow.

The simplest setup is enough to validate the product: upload policy documents, import historical tickets and orders, then review the generated drafts and summaries.

Setup philosophy Roll out SparkWren the same way the rest of the site describes it
Start Documents + exports No deep integration required on day one.
Focus One workflow Pick one queue and one approval model first.
Review Human-in-the-loop Validate output quality before expansion.
Expand Adjacent workflows Grow only after the first use case is trusted.
1. Upload your source material

Return policy, shipping policy, warranty logic, FAQ, and macros.

2. Import queue samples

Bring in CSV or inbox exports plus order and tracking context.

3. Define review rules

Choose when managers should approve: refunds, VIPs, legal tone, compensation, or fraud risk.

4. Run the workflow

Receive draft replies, internal notes, issue labels, and next actions.

5. Publish team review views

Share queue summaries and edge-case reports with leads and operators.

Start here

Choose one workflow and prove it before you broaden the rollout.

What you upload first

  • Shipping, return, refund, and warranty policies
  • FAQ pages and saved reply language
  • Historical ticket samples with outcomes
  • Order exports and tracking snapshots for context-heavy workflows

What your team reviews first

  • Draft accuracy against policy
  • Whether risk flags and approvals make sense
  • Whether internal notes are useful to ops or managers
  • Whether the workflow is saving enough time to justify expansion
Workflow docs map

Each documentation path should connect back to a real workflow page.

Wrong item workflow

Focus on order verification, evidence collection, and policy-backed next steps.

Agent QA workflow

Focus on batch review, coaching signals, and consistency across brands or queues.

Policy drift workflow

Focus on keeping macros, SOPs, and live queue behavior aligned after policy changes.