Product flow & architecture
This page maps the journey a user takes across the three DataChord suites and shows how the platform's objects flow from one stage to the next. If you read only one page of these docs before opening the product, read this one.
The DataChord journey
At each step, DataChord produces a named object that the next step consumes. Nothing is implicit:
| Step | You produce | The next step consumes |
|---|---|---|
| StructScan | SourceProfile | MapCraft uses the profile to bootstrap the source schema |
| MapCraft | Project with confirmed MappingFields, Rules, Lookups | FlowBridge compiles a DAGDocument from them |
| FlowBridge | DAGDocument + DAGRuns + PackageArtifacts | Your warehouse, partner, or runtime |
The platform surface
The three suites share a single workspace and a single tenancy model.
- Auth and tenancy. Every user signs in via JWT and is bound to an organisation. Every object (profile, project, DAG, run, package) is scoped to that organisation; you can never see another org's data.
- Activity Feed. The Dashboard surfaces a unified feed of recent StructScan profiles, MapCraft projects, and FlowBridge runs so you can pick up where you left off.
- Contracts. Every shared object shape is defined by a JSON Schema checked into the platform. You can validate exports against the same contracts the server uses.
Object hand-off model
The objects below are the canonical building blocks. You will see each one named in the walkthroughs.
| Object | Produced by | Holds | Lives in |
|---|---|---|---|
UploadedFile | StructScan upload | File metadata, storage path | StructScan |
SourceProfile | StructScan profiling | field_profiles, quality_summary, ai_insights | StructScan |
Project | MapCraft create / Push from StructScan | Lane, status, source & target schema refs | MapCraft |
Schema | MapCraft schema upload (or auto from profile) | canonical_fields (JSON) | MapCraft |
Mapping / MappingField | MapCraft canvas / table / auto-map | Source path → target path, status, confidence | MapCraft |
TransformationRule | MapCraft rules editor | Natural language → Python / SQL / JSONata | MapCraft |
LookupTable | MapCraft lookups | Key column, value column, rows | MapCraft |
Version | MapCraft save version | Project snapshot, change summary | MapCraft |
DAGDocument | MapCraft DAG view (compile) | Nodes, edges, metadata (the IR) | FlowBridge |
DAGRun | FlowBridge trigger / schedule | Status, trigger, records_processed, errors | FlowBridge |
DeliveryAdapter | FlowBridge Deliver | Adapter type (SFTP/HTTP/FILE/WEBHOOK), config | FlowBridge |
Policy | FlowBridge Policy-as-Code | Mask / tokenize / redact rules per field | FlowBridge |
PackageArtifact | FlowBridge Package | Lambda / Cloud Run / Cloud Function bundle | FlowBridge |
note
Two practical consequences of the hand-off model:
- You can start at any stage. If you already have a target schema and a source file, you can skip StructScan and create a MapCraft Project directly. If you already have a compiled DAG, you can schedule it in FlowBridge without opening MapCraft again.
- Each step is reproducible from its predecessor. Re-running StructScan
on a refreshed file gives you a new
SourceProfile, not an in-place edit. Re-compiling a MapCraft project gives you a newDAGDocumentversion. History is the default.
Status lifecycles at a glance
API surface at a glance
The product is a single UI, but it is backed by clearly-prefixed APIs. You do not need to know these to use DataChord — they are listed here so the glossary feels less magical.
Top-level API prefixes (click to expand)
| Prefix | What it powers |
|---|---|
/api/v1/auth | Sign-in, session, org context |
/api/v1/structscan | Upload, profile, raw preview, push-to-MapCraft |
/api/v1/mapcraft | Projects, schemas, mappings, rules, lookups, versions, export |
/api/v1/execution | DAG compile, trigger, runs, schedules, preview SQL |
/api/v1/flowbridge/extract-load | Warehouse extract / load + CDC config |
/api/v1/flowbridge/deliver | Delivery adapters (SFTP / HTTP / FILE / WEBHOOK) |
/api/v1/flowbridge/sync | OAuth grants, sync-to-app targets |
/api/v1/flowbridge/package | Serverless build / publish |
/api/v1/policy | Field-level masking / tokenizing / redacting |
/api/v1/drift | Drift events, fix proposals |
/api/v1/activity | Unified Dashboard activity feed |
Where to go next
Open the walkthrough for the suite you want to start with. The three
walkthroughs share a single sample dataset (customer_orders.csv) and read
naturally end-to-end:
- StructScan walkthrough — upload the file, get a profile, push to MapCraft.
- MapCraft walkthrough — confirm mappings, add a lookup and a rule, compile the DAG.
- FlowBridge walkthrough — apply a policy, schedule it, deliver it, package it.