Filing → Configured Enrollment

Unstructured data to structured data to fully configured.

The pattern applies to contracts, RFPs, SOWs, work orders, or any inputs that produce a defined set of outputs. Here is one example. Requirements are buried in sets of up to 20 .pdf files. Every set is different. This workflow extracts those requirements and automates implementation end-to-end. Every decision is captured. The process is 100% auditable.

.md instruction files
control agents
Python (LangGraph)
controls orchestration
Python
does deterministic work
PostgreSQL
receives data output

Quality loops diagnose, resolve, and hand off to a human in the loop when an issue cannot be resolved. The diagnosis is provided and the human is given resolution options.

Sample filings · public SERFF records click ↓ to download · drag a tile onto the zone below to start
Loading samples…
Drag and drop a sample to start workflow
Run
no active run
Filing
·
Started
·
Idle
Workflow awaiting run
  1. Classify
    documents are categorized
  2. Diagnose
    filing completeness assessed
  3. Extract
    structured product produced
  4. Transform
    code · 13 tables
  5. Write
    code · ON CONFLICT upserts
  6. Validate
    LLM · DB compared to filing
  7. Reconcile
    human · only when validation flags
Product · being assembled
Filing
·
Carrier
·
Type
·
Plans
·
Records
·
Validation
·
Classify
Diagnose
Extract
Transform
Write
Validate
Reconcile
Each tab above is a workflow step. While a run is active, this panel shows the live mechanism of the current step: an audit block naming the actor (LLM, code, or human), the model and instruction file in use, then below it the prompt being assembled, the tokens streaming back as the model produces structured JSON, or the SQL inserts executing as code steps write rows to PostgreSQL. After the run, every tab is clickable for review.
Enrollment Tool

Where the configured product becomes a sellable application

This panel will display the consumer-facing enrollment application as soon as a filing has been successfully loaded and configured. The eApp reads from the same database the pipeline writes to. Same data, different audience. Drop a sample to see it appear.

Waiting for a successful run.
Event ledger no events yet ▾ collapse
Events will appear here as the workflow runs. Newest at the bottom.