You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
An intelligent, voice-native debt collection platform powered by LiveKit, Groq, and Deepgram. Features real-time negotiation, sub-500ms latency, FDCPA compliance guardrails, and a cinematic command center UI.
AI-driven debt collection platform for FedEx that automates case allocation, tracks every customer interaction in real-time, and provides complete visibility into DCA operations. Built with Flask, React, and local LLMs.
ML-powered desktop app for simulating debt collection portfolio strategies. Blends CatBoost and LightGBM models to predict recovery outcomes across Low / Medium / High effort scenarios. Built with Python + Tkinter GUI, trained on historical collection data.
Debt Validation Prompter — split-pane letter generator with local templates + a JSON prompt pack for optional AI polish (no API calls baked in). // built with a little agentic assistance — local-first, so there are no “hidden API surprises”
Air-gapped dispute-response pipeline for collection law firms. Drafts letters, checks every claim against the source ledger, and stages all output for human review. Runs fully offline on a local model. Built for Brainstorm 2026.
Recoup AI — self-improving, FDCPA-compliant AI voice agents for debt collection with a Salesforce-style CRM. Pipecat + NVIDIA Nemotron + Cekura (YC Voice Agents Hackathon).
Seeded Python generator for a synthetic debt collection agency data set: 10,000 accounts, ~200,000 activity notes, and 62 documented data quality defects for hands-on AI training sessions.
Design decisions for Wax, an event-sourced .NET framework, and HiveAR, the free, open-source accounts receivable and debt collection platform being built on it by The OpenAR Collective, a vendor-neutral nonprofit. Read them, question them, and contribute what you know about how collections work; no coding required. Start with START-HERE.md.
Model Context Protocol server for the Payt API. Lets Cursor, Claude Desktop and other MCP clients call Payt's debtor, invoice, payment and credit-case endpoints as tools.
Model Context Protocol (MCP) server for dunning letters and overdue invoice payment reminders: reminder, final notice and aging. Chase overdue invoices on a ladder anchored to the due date: reminder 1, reminder 2, the final notice, with the aging and the day's chase list. Works with Claude Desktop, Claude Code and Cursor.