
AutoOps is a fully autonomous financial operations platform designed to automate invoice processing, policy enforcement, and USDC payments using Circle’s Arc L1, Circle Payment APIs, and a multi‑system architecture. Our system consists of four coordinated components: AI Agent (Decision Engine): A deterministic rules-based agent evaluates invoices against company financial policies. It outputs strict, structured JSON decisions (APPROVE, NEGOTIATE, REJECT), ensuring zero‑hallucination financial logic and enforcing a finance‑first, risk‑averse approach. Frontend Dashboard: A production‑ready UI built for finance teams to upload invoices, view decisions, track vendors, and monitor transactions in real time. Users can view invoice history, agent decisions, and blockchain transaction receipts from Arc. Blockchain & Payments Layer: Smart contracts deployed on Circle’s Arc Testnet manage vendors, invoices, and fund release. The backend integrates Circle’s Wallets API and Payments API to create wallets, initiate USDC transfers, and confirm settlements. Contracts include: VendorRegistry InvoiceManager USDC handling + escrow lifecycle Backend + Supabase Integration: A unified FastAPI/Node backend orchestrates: Agent → Contract → Circle API workflows Supabase logging for audit trails Webhooks for payment status updates Secure routing between systems 1–3 Together, AutoOps automates the entire finance operations pipeline—from invoice upload to blockchain‑settled payment—without human intervention. The system is built for scalability, enterprise reliability, and real‑world financial auditability.
24 Jan 2026

SWASTIK (Smart Workflow AI Support Triage & Knowledge) is an intelligent multi-agent orchestration platform designed to modernize customer support operations using AI-driven automation. Built on IBM watsonx Orchestrate, this system utilizes a coordinated team of autonomous agents to analyze, classify, resolve, and manage customer issues in real time. The platform includes a Supervisor Copilot that orchestrates three specialized agents: a Triage Agent to classify urgency and sentiment, a Knowledge Agent to retrieve policy-based solutions, and an Actions Agent that simulates ticket handling logic. These agents work together to generate structured insights such as issue category, priority, sentiment, suggested replies, and ticket recommendations. SWASTIK features a modern, dark-themed chat interface inspired by ChatGPT, delivering a seamless user experience for support teams. The system reduces manual effort, improves response speed, and ensures consistent, accurate customer communication across billing, technical, and account-related concerns. This project demonstrates how agentic AI can transform enterprise customer service workflows through orchestration, automation, and intelligent decision-making.
23 Nov 2025