Skip to main content
fh farhanhelmy/code

Sidegent Deep Agent Builder

active
AI/ML SaaS Agents RAG Analytics

Overview

Sidegent Deep Agent Builder is a multi-tenant platform for teams that need controlled internal assistants and customer-facing chat experiences grounded in their own knowledge.

The product focuses on governance, usage visibility, organization management, and operational control rather than one-off chatbot demos.

Key Features

  • Assistant builder and runtime controls: Configure behavior, tool access, settings, and prompt refinement workflows from a web dashboard
  • Knowledge base ingestion and retrieval: Upload files, chunk and index documents, and ground answers using retrieval flows designed for internal content
  • Embeddable chat: Deploy assistants as embedded chat widgets for external users with conversation tracking and analytics
  • Organization analytics and admin tooling: Monitor agent usage, token consumption, embed activity, and organization-level performance from admin dashboards
  • Billing architecture: Support budget controls and a prepaid wallet model with ledger-style settlement, low-balance thresholds, and org-specific billing visibility
  • Flexible auth: Email/password, magic-link sign-in, invitation-based onboarding, and Google sign-in for platform access

Technology Stack

  • Frontend: React 19, TypeScript, Vite, TanStack Router, Tailwind CSS, shadcn/ui
  • Backend: FastAPI, SQLModel, Alembic, PostgreSQL, pgvector
  • Workflow intelligence: Managed LLM runtime, retrieval, embeddings, web research tools, and usage controls
  • Infrastructure: Nx monorepo, AWS, Resend, Grafana, Loki

Engineering Notes

This was built as a real product surface rather than a prototype. The work spans multi-tenant organization management, background knowledge-base jobs, runtime usage tracking, embed analytics, and the accounting layer needed to turn raw model usage into customer-facing billing.

Outcome

The platform turns knowledge workflows into a controllable SaaS product. It gives teams a way to ship internal assistants and customer-facing chat experiences while keeping knowledge, access, analytics, and spend visible in one system.