piyush-nst/resolveai

How piyush-nst/resolveai works

ResolveAI supports two user-facing flows: a knowledge copilot that retrieves support evidence and returns cited answers, and a resolution agent that handles order actions, approval pauses, and escalation. The graph also preserves the documented ingestion/indexing, mock order system, and evaluation capabilities. Relationships are limited to calls or connections shown in excerpts, the source index, or README; unsampled subsystems are represented at directory level where needed.

Main entry points: retrieval/pipeline.py Retrieval Pipeline · main.py FastAPI Endpoints · graph.py Agent Workflow · ingestion/pipeline.py Index Builder

Components 26

User Interface

Support Web UI src/resolveai/webui/static/app.js
FastAPI Endpoints src/resolveai/api/main.py

Resolution Agent

Agent Workflow src/resolveai/agent/graph.py
Agent Runtime src/resolveai/agent/runtime.py
Agent Tools src/resolveai/agent/tools.py

Knowledge Retrieval

Retrieval Pipeline src/resolveai/retrieval/pipeline.py
Semantic Search src/resolveai/retrieval/semantic.py
BM25 Search src/resolveai/retrieval/bm25.py
Hybrid Fusion src/resolveai/retrieval/hybrid.py
Cross-Encoder Reranker src/resolveai/retrieval/reranker.py
Index Builder src/resolveai/ingestion/pipeline.py
Support Data Loader src/resolveai/ingestion/loader.py
Document Chunker src/resolveai/ingestion/chunker.py
Embedding and Chroma Index src/resolveai/ingestion/vectorstore.py

Order Operations

Order System Client src/resolveai/orders/client.py
Order Database src/resolveai/orders/db.py
Order Records src/resolveai/orders/models.py
Demo Data Seeder src/resolveai/orders/seed.py

Model Services

Agent LLM Client src/resolveai/llm/client.py
Groq LLM
Gemini Fallback
Grounded Answer Generation src/resolveai/generation/__init__.py
Configuration src/resolveai/config.py

Evaluation

Retrieval Evaluation src/resolveai/evaluation/runner.py
Agent Scenario Evaluation evals/runner.py

Other components

Customer

Connections 29

Customer → Support Web UI: uses
Customer → FastAPI Endpoints: submits queries
Support Web UI → FastAPI Endpoints: calls endpoints
FastAPI Endpoints → Retrieval Pipeline: runs retrieval src/resolveai/api/main.py
FastAPI Endpoints → Grounded Answer Generation: generates answer src/resolveai/api/main.py
FastAPI Endpoints → Agent Runtime: runs chat and approvals src/resolveai/api/main.py
Agent Runtime → Agent Workflow: builds workflow src/resolveai/agent/runtime.py
Agent Workflow → Agent LLM Client: interprets requests src/resolveai/agent/graph.py
Agent Workflow → Agent Tools: dispatches actions src/resolveai/agent/graph.py
Agent Workflow → Retrieval Pipeline: searches knowledge src/resolveai/agent/graph.py
Agent Tools → Order System Client: performs order actions src/resolveai/agent/tools.py
Agent Runtime → Order Database: configures persistence src/resolveai/agent/runtime.py
Order System Client → Order Database: reads and writes src/resolveai/orders/client.py
Order Database → Order Records: maps records src/resolveai/orders/db.py
Demo Data Seeder → Order Records: creates demo records src/resolveai/orders/seed.py
Index Builder → Support Data Loader: loads support data src/resolveai/ingestion/pipeline.py
Index Builder → Document Chunker: chunks documents src/resolveai/ingestion/pipeline.py
Index Builder → Embedding and Chroma Index: embeds and indexes src/resolveai/ingestion/pipeline.py
Retrieval Pipeline → Semantic Search: runs semantic search src/resolveai/retrieval/pipeline.py
Retrieval Pipeline → BM25 Search: runs keyword search src/resolveai/retrieval/pipeline.py
Retrieval Pipeline → Hybrid Fusion: fuses candidates src/resolveai/retrieval/pipeline.py
Retrieval Pipeline → Cross-Encoder Reranker: reranks results src/resolveai/retrieval/pipeline.py
Semantic Search → Embedding and Chroma Index: queries vectors src/resolveai/retrieval/semantic.py
Agent LLM Client → Groq LLM: calls primary model src/resolveai/llm/client.py
Agent LLM Client → Gemini Fallback: fails over src/resolveai/llm/client.py
Grounded Answer Generation → Groq LLM: generates response src/resolveai/generation/__init__.py
Retrieval Evaluation → Retrieval Pipeline: evaluates retrieval src/resolveai/evaluation/runner.py
Agent Scenario Evaluation → Agent Workflow: runs scenarios evals/runner.py
Agent Scenario Evaluation → Order System Client: checks order outcomes evals/runner.py

Drawn by GitDiagram on November 10, 2026 from the repository's file tree, README and source files. Read as Markdown ·