Turn EDI, JSON & XML into AI-Ready Snowflake Data for Government

Public sector agencies hold some of the most valuable data in the country, and much of it arrives in the hardest formats to use. Medicaid enrollment and claims come in as X12 EDI. Pharmacy transactions come in as NCPDP. Clinical exchange runs on HL7, FHIR, and CDA. Justice, public safety, and human services systems exchange information as NIEM-conformant XML and JSON. Change data capture pipelines from Kafka and Salesforce deliver a constant stream of nested JSON and Avro events, and document databases such as MongoDB store records whose structure varies from one document to the next. Before any of it can support reporting, analytics, or AI, it has to be transformed into accurate and complete relational tables. The conventional route is a long data integration project built with a combination of tools and hand-written mappings that cover only the structures someone documented in advance.

DataPancake, from TDAA Technologies Corp., takes a discovery-first approach. It is a patented Snowflake Native App that scans the full population of data an agency makes available to it, combines what it finds with licensed standards libraries such as X12, and lets the agency's own team shape the result through configurable metadata. DataPancake then generates the SQL that creates governed relational dynamic tables and views in Snowflake. The process is deterministic, not sampled or inferred by a language model, so the same input produces the same output and the result can be explained to an auditor.

For government buyers, the architecture matters as much as the output. DataPancake installs and runs entirely inside the agency's own Snowflake account. TDAA does not host agency data, and the application makes no external API calls. It is in use today in a Snowflake government region. The generated code is standard Snowflake SQL that staff can inspect before deployment, and once deployed it runs and refreshes without DataPancake, so agencies avoid a permanent runtime dependency on a vendor. Security policy metadata can be carried into the generated objects, helping ensure downstream users see only the data they are authorized to see.

Agencies also get evidence, not assumptions. Reconciliation metrics compare raw source data to the materialized tables, and schema drift detection alerts teams when new structures appear, which matters most for event streams and document stores where the structure changes without notice. Pricing is based on structural complexity or named standards libraries, not data volume, so the annual cost is known before an order is placed and fits fixed appropriations.

Faster and more accurate access to this data changes what an agency can do with it. State Medicaid programs can validate the encounter and enrollment data submitted by managed care organizations instead of accepting it as reported. They can calculate HEDIS and other quality measures internally, while there is still time to act on the results, instead of waiting for measures that arrive long after services were delivered. Program integrity teams can bring complete claims, provider, and pharmacy data into fraud, waste, and abuse analytics, and surface the eligibility and coordination of benefits gaps that lead to improper payments. Federal reporting such as T-MSIS can draw on data whose completeness has been measured, not assumed. The same foundation serves agencies outside healthcare, from justice and public safety information sharing to AI initiatives that depend on trustworthy inputs. All of it is handled in one application, by a small team, in days instead of months.

DataPancake is available to public sector customers through Carahsoft and its contract vehicles, including NASPO ValuePoint.

Frequently Asked Questions