## Models:
- `testConfig`
	- S3 Path: /tests/definitions/{testId}.json
	- Properties:
		- id: string
		- description: string
		- documentSet: string
		- workflow_deployment_name: string //from vellum
		- release_tag: string //from vellum
		- lastRun: { 
			- path //s3 path to the run
			- dateTime //the date and time the ran began
			- latency //ms it took to run
			- cost //cost, to 4 decimal places
			- accuracy //decimal from 0 to 1 representing the number of correct responses }
- testResult
	- id: string //8 digit base 62 random string
	- dateTime: //date time the test started
	- documentSetId: //the id of the document set that was run
	- documents: //array
		- documentId //string
		- expectedDocument: json object that should match the document
		- responseDocument: json object //the document part of the response
		- responseMetadata: json object //the metadata part of the response
		- incorrectFields
			- jsonPath
			- expected
			- actual
		- success //bool - only if there are no incorrect fields
		- error
		- usage
		- cost
		- latency
	- usage
	- cost
	- latency
	- accuracy
	- error: string
- testResultSummary
	- id: testResult.id
	- datetime: testResult.dateTime
	- usage: testResult.usage
	- cost: testResult.cost
	- latency: testResult.latency
	- accuracy: testResult.accuracy
	- error: testResult.error
	- documentSetId: testResult.documentSetId
- testGroup
	- id: string
	- description: string
	- testIds: string[]
	- lastRun: string
- testDocumentSet
	- id: string
	- description: string
	- documentIds: string
		- id: string
		- expectedDocument: json object to compare the result of the test to
- documentActual
- 

## S3 Paths
- /tests/definitions/{testId}.json
- /tests/groups/{testGroupId}.json
- /tests/definitions-archived/{testId}.json
	- When a test is archived, move it here
- /tests/results/{testId}/{yyyyMMddHHmmss}/{id}.json - Contains a `testResult` object. The date/time is for when that group of tests (or single test) was started.
- /tests/results/{testId}/{yyyyMMddHHmmss}/{id}-summary.json - a smaller summary of the `testResult` - as a `testResultSummary`
- /documents/{id}
- /documents-processed/{id}
- /documents/test-sets/{setId}.json
- /documents-actuals/{id} - contains the expected json result of the document

## APIs:
- 
- POST /api/ai-tools/process-document
	- JSON BODY
		- { "document-s3-key": "path/on/s3.pdf" }
	- RETURNS
		- Result
		- Cost
		- Usage
- run a test - POST /api/ai-tools/test/{test-id}/run
	- No body
	- Loads the given testConfig from S3, and calls "process-document" for every document in the test config(parallelize to process 10 at a time)
	- Updates the testConfig.json with the path to the test-results
- create a test  - POST /api/ai-tools/create-test
- list  tests - GET /api/ai-tools/list-tests?isArchived=true|false
	- Lists all the testConfigs from S3
	- if isArchived=true, only show the archived tests
	- if isArchived=false, do not show the archived tests
- get test result - GET /api/ai-tools/test/{test-id}/result?timestamp={testResult:yyyyMMddHHmmss}&id={testResultId}
- list test result summaries - GET /api/ai-tools/test/{test-id}/result-summaries
	- Grabs all the testSummaryResult entries for this test
- list document sets: returns the name and description of all the document sets. this should be cached on the server
- create document set
- edit document set
- update document set expected value

## Screens:
 - Login: `/login` use hardcoded user and password
 - Dashboard: `/dashboard`
	 - Lists all the testConfigs
		 - Click a toggle to show the archived tests
		 - Show: 
			 - Name
			 - Description
			 - Last Run date / cost / accuracy
		 - Click the test config to go to the test screen
	 - Button to create test
		 - Show modal
		 - Enter Id (textbox) and Description (textarea)
		 - Choose a document set (dropdown)
		 - Enter a deployment name
		 - Enter a deployment tag
		 - Create the test
		 - Refresh the dashboard to include the test
	 - Can select multiple tests to create a test group
	 - Lists the test groups - a test group is a list of tests you can run all at once
 - Test Details `/test/{testId}`
	 - Name and description of the test
	 - List of recent test executions. Group by testDocumentSet
		 - date/time
		 - cost
		 - accuracy
		 - documentGroup
	 - Click a button to run the test
		 - Choose a testDocumentSet to run it on
 - Test Results: `/test/{testId}/{testResultDate:yyyyMMddHHmmss}/{testResultId}`
	 - By default, only show the documents where success = false (there is at least one incorrect field)
		 - Can click a toggle to show all the documents
	 - Split pane view. When you click on a document in the left pane, show the document in the right pane (collapsed by default). 
	 - For each document, show the expected and actual values for each field (show only mismatches by default, can click to show all). Click on a field's "expected" value to modify it inline. Automatically save the updated value in the document set