Skip to content

saagpatel/AssistSupport

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

241 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

AssistSupport

Rust License Platform

Your support team's second brain — ML-powered answers from your own knowledge base, in under 25ms, without sending a single query to the cloud.

AssistSupport combines local LLM inference with a hybrid ML search pipeline to generate accurate, KB-informed IT support responses. A logistic regression intent classifier (85.7% accuracy) routes queries before a TF-IDF retriever finds candidates, and a cross-encoder reranker (ms-marco-MiniLM-L-6-v2) sharpens relevance before the response is drafted. The entire pipeline — app, sidecar, and model inference — runs on your machine. Core workspace data is encrypted at rest.

User asks:    "Can I use a flash drive?"
ML Intent:    POLICY detected (86% confidence)
Search finds: USB/removable media policy in 21ms
Reranker:     Cross-encoder confirms top result relevance
AI drafts:    "Per IT Security Policy 4.2..."
You copy:     Paste into Jira — done in under a minute

Features

  • ML Intent Classification — Logistic regression classifier routes queries to the right search strategy before retrieval even starts
  • Sub-25ms Hybrid Search — p50: 8ms, p95: 82ms across 3,500+ KB articles; TF-IDF + cross-encoder reranker pipeline
  • Encrypted Local Workspace — Core SQLite database and stored secrets are protected with wrapped keys and encrypted-at-rest storage; no cloud dependency for the primary workflow
  • Trust-Gated Responses — Confidence modes (answer / clarify / abstain), claim grounding map, citation-aware copy safety for low-confidence output
  • Self-Improving Feedback Loop — KB gap detector surfaces repeated low-confidence topics and tracks remediation over time
  • Ops-Ready Workspace — Deployment preflight, rollback flows, eval harness runs, triage clustering, and runbook sessions built in

Quick Start

Prerequisites

  • Node.js 20+
  • pnpm 9+
  • Rust toolchain (stable) + Tauri v2 prerequisites for macOS

Installation

git clone https://github.com/saagpatel/AssistSupport.git
cd AssistSupport
pnpm install
cp .env.example .env

Run (development)

pnpm dev

Build (desktop app)

pnpm tauri build

Tech Stack

Layer Technology
Desktop shell Tauri 2 + Rust
Frontend React + TypeScript + Vite
ML search TF-IDF, Logistic Regression, ms-marco-MiniLM-L-6-v2
Local storage SQLite (encrypted)
LLM inference Local via Ollama (optional)
Fonts IBM Plex Sans, JetBrains Mono

Architecture

AssistSupport is a Tauri 2 desktop app with a Rust backend handling search, encryption, and LLM orchestration. The ML pipeline runs as a local sidecar: intent classification happens first, then candidate retrieval via TF-IDF index, then cross-encoder reranking to select the most relevant KB articles before response generation. The feedback loop writes ratings back to a local SQLite store and periodically surfaces gap analysis via the Ops workspace.

License

MIT

About

Local-first IT support assistant — ML intent classification, sub-25ms search, encrypted workspace, zero cloud dependency

Topics

Resources

License

Code of conduct

Contributing

Security policy

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors