How yarram-koushik/insightforgeai works
InsightForgeAI is a business-intelligence assistant that turns uploaded or connected company data and user questions into grounded answers, analytics, and shareable reports. The main workflow runs from the Streamlit UI through ingestion and workspace data into conversational context, intent routing, and specialized agent work; the API provides a second user-facing boundary. Separate product branches cover knowledge-grounded answers, proactive scans, dashboards and exports, and scheduled delivery. The source excerpts confirm selected calls, but not all UI/API wiring or feature internals; uncertain links are omitted.
Main entry points: orchestrator.py Agent orchestrator · app.py Streamlit UI · scheduling.py Schedules and delivery · main.py FastAPI boundary
Components 22
User experience
- Streamlit UI
app/frontend/app.py - FastAPI boundary
app/backend/main.py
Question and analytics
- Agent orchestrator
app/agents/orchestrator.py - Conversation context
app/core/context_memory.py - Intent router
app/agents/router.py - Data query agent
app/agents/sql_agent.py - Analytics agent
app/agents/analytics_agent.py - Insight agent
app/agents/insight_agent.py - Forecast agent
app/agents/forecast_agent.py
Data and meaning
- File ingestion
app/core/ingestion.py - Workspace data
app/core/data_manager.py - SQL connectors
app/core/connectors
Enterprise features
- Knowledge base
app/core/knowledge_base.py - Proactive insights
app/core/proactive.py - Dashboards
app/core/dashboard.py - Report export
app/core/export.py - Schedules and delivery
app/core/scheduling.py
Other components
- Business user
- API client
- Company databases
- Slack or email
- LLM providers
Connections 22
- Business user → Streamlit UI: asks or uploads
- API client → FastAPI boundary: sends requests
- Streamlit UI → File ingestion: reads files
- File ingestion → Workspace data: loads datasets
- Streamlit UI → Workspace data: saves datasets
- Streamlit UI → Agent orchestrator: runs questions
- FastAPI boundary → Agent orchestrator: runs questions
- Agent orchestrator → Conversation context: expands follow-ups
- Agent orchestrator → Intent router: classifies intent
- Intent router → LLM providers: requests classification
- Agent orchestrator → Data query agent: dispatches data queries
- Agent orchestrator → Analytics agent: dispatches analytics
- Agent orchestrator → Insight agent: dispatches insights
- Agent orchestrator → Forecast agent: dispatches forecasts
- Agent orchestrator → Knowledge base: requests grounded answers
- Agent orchestrator → Proactive insights: requests scans
- Streamlit UI → Proactive insights: scans selected table
- SQL connectors → Company databases: reads schema and tables
- Streamlit UI → Dashboards: pins widgets
- Streamlit UI → Report export: exports reports
- Schedules and delivery → Agent orchestrator: runs scheduled questions
- Schedules and delivery → Slack or email: delivers results