Skip to content

Dashboard ​

Table of contents ​

Purpose ​

The Streamlit dashboard presents retention signals, evidence, validation reports, and advisor output.

Application structure ​

text
apps/dashboard/streamlit-app/
services/dashboard/Dockerfile
.streamlit/

Reusable layout and rendering concerns should remain separate from data access and backend communication.

Runtime contract ​

text
RETAINAI_DASHBOARD_MODE=api
RETAINAI_API_PROVIDER=aws-lambda
RETAINAI_API_URL=https://api.retainai.hubertronald.dev
RETAINAI_BACKEND_TOKEN=<Secret Manager reference>
RETAINAI_AI_PROVIDER=disabled
RETAINAI_ENABLE_RAG=false
RETAINAI_VECTOR_STORE=disabled

Application footer:

text
Designed and built with ♥ and AI assistance by Hubert Ronald
© 2026 RetainAI — All rights reserved

RetainAI is attributed directly to Hubert Ronald. Liasoft remains a separate creative brand.

Container build ​

Canonical Dockerfile:

text
services/dashboard/Dockerfile

Build context:

text
repository root

Cloud Run deployment ​

bash
IMAGE_TAG="<immutable-tag>"   ./scripts/deploy_dashboard_cloud_run.sh

The script builds, pushes, and deploys a new revision without applying Terraform.

Scaling ​

text
minimum instances = 0
service maximum = 1
revision maximum = 1

Every deployment must preserve those values.

Validation ​

bash
curl -sSIL https://retainai.hubertronald.dev
bash
gcloud run services describe   retainai-dashboard   --project=coplayground   --region=us-east4   --format=json

Internationalization readiness ​

User-facing text should move into locale catalogs before v1.0:

text
locales/en.json
locales/es.json

Rendering components should consume stable message keys rather than hard-coded English or Spanish text.

Dashboard contract ​

The RetainAI dashboard consolidates the product-facing views and the runtime integration contract in one place.

Application pages ​

The dashboard exposes four primary product pages:

  1. Executive Overview
    Executive-level retention indicators, risk summaries, and decision context.

  2. Prediction Center
    Employee-level prediction workflows and model-assisted retention analysis.

  3. Explainability Explorer
    SHAP-based feature attribution and explanation views for model decisions.

  4. Survival Analytics
    Time-to-event and retention-horizon analysis.

API domains ​

The dashboard consumes the branded RetainAI backend through these API domains:

text
/api/v1/overview
/api/v1/predict
/api/v1/explainability
/api/v1/survival

The production API base URL is:

text
https://api.retainai.hubertronald.dev

Runtime modes ​

The dashboard contract recognizes four runtime modes:

ModePurpose
localLocal development using local files and local execution.
apiDashboard integration with the deployed RetainAI API.
s3Data or artifact access through Amazon S3-backed workflows.
awsAWS-oriented runtime integrations, including Lambda and related services.

The currently deployed dashboard uses the api mode against the branded backend. The local, s3, and aws modes remain documented compatibility and evolution paths for development, artifacts, and cloud integrations.