Collections Optimizer

Collections Optimizer

A daily collections work-queue for banks, NBFCs, MFIs, BNPL, telecom and utilities. Scores propensity-to-pay and self-cure, ranks accounts by expected recovery uplift from contact (the moveable middle, not oldest or biggest first), recommends channel / timing / tone, and assigns work to agents. Contact-hour and frequency caps are hard constraints.

Registry name	ai.saifs/collections-optimizer
Endpoint	https://collections-optimizer-mcp.saifs.ai/mcp
Transport	Streamable HTTP (remote, nothing to install)
Auth	OAuth 2.1 — PKCE (S256) and dynamic client registration; your client opens a sign-in page on first use
Publisher	SAIFS AI

Install

VS Code / GitHub Copilot

Add to .vscode/mcp.json (workspace) or your user mcp.json:

{
  "servers": {
    "saifs-collections-optimizer": {
      "type": "http",
      "url": "https://collections-optimizer-mcp.saifs.ai/mcp"
    }
  }
}

Then start the server from the MCP view and sign in when prompted.

Claude (claude.ai, Desktop)

Settings → Connectors → Add custom connector, and paste:

https://collections-optimizer-mcp.saifs.ai/mcp

Claude Code

claude mcp add --transport http saifs-collections-optimizer https://collections-optimizer-mcp.saifs.ai/mcp

Any other MCP client

Point it at https://collections-optimizer-mcp.saifs.ai/mcp using the Streamable HTTP transport. The server returns 401 with a WWW-Authenticate: Bearer … resource_metadata=… challenge, so OAuth-aware clients sign in automatically.

Tools

Tool	What it does
score_propensity	P(will pay), P(self-cure) and expected recovery for an account.
prioritize_queue	Daily work-list ranked by expected recovery uplift from contact.
recommend_contact_strategy	Best channel, timing and tone-tier (call / WhatsApp / notice) within fair-collection hours.
flag_self_cure	Accounts likely to pay without contact — skip to avoid wasted or excess outreach.
predict_roll_rate	Survival-based bucket-to-bucket roll / cure forecast for provisioning.
optimize_agent_allocation	Assign accounts to agents by skill and success history (Hungarian method).
explain_priority	Why this account and this strategy — for agent trust and audit.
watch_queue	Nightly re-score so the morning queue is fresh.

Example prompts

• “Here’s our overdue book. Who should agents work today, and who will likely self-cure?”
• “For account A-1042, what’s the best channel and time to contact, within permitted hours?”
• “Allocate this queue across 8 agents by who recovers which product best.”

Support

• Website: https://saifs.ai
• Issues: https://github.com/Saify-Technologies/saifs-ai-mcps-docs/issues
• Privacy: https://saifs.ai/privacy_policy