Inspiration

The earliest signs of customer churn rarely appear first in CRM. They show up inside Slack as scattered signals: a support escalation, a sales concern, an incident note, an engineering ownership gap, or an executive update that arrives too late.

RescueOps turns Slack into an early-warning and rescue system for revenue risk.

What it does

RescueOps is a Slack-native revenue rescue agent that detects hidden customer revenue risk, explains the evidence, and launches an approved rescue workflow inside Slack.

In the demo, RescueOps scans Acme Robotics using Slack Real-Time Search and MCP business context, then identifies:

  • USD 500,000 in revenue at risk
  • 10 evidence signals across live Slack Real-Time Search results and connected MCP business context
  • 100% critical risk
  • early warning signals before the risk is visible in CRM
  • diagnosis reduced from 4 hours to 45 seconds
  • rescue plan time reduced from 1 day to 2 minutes
  • USD 245,000 expected protected revenue

How we built it

RescueOps is built as a Slack app using Socket Mode, slash commands, Block Kit, interactive buttons, Slack Web API, Slack AI assistant hooks, Slack Real-Time Search, and MCP server integration.

The live proof path is:

/rescueops rts-check -> /rescueops live acme -> Live Slack RTS + MCP business context

The core engine converts evidence into a RescueCase:

Evidence -> risk score -> root causes -> owner recommendations -> rescue workflow -> impact receipt

Slack Real-Time Search provides live workspace evidence. MCP business context enriches that evidence with CRM, support, incident, and ownership signals. The same pipeline then generates the score, evidence explanation, rescue actions, owner assignments, and impact receipt.

A deterministic demo fallback exists only for reproducible testing when Slack credentials or RTS access are unavailable. An optional Qwen3 open-source reasoning layer can rewrite rescue communication from grounded RescueCase JSON without replacing the deterministic scoring engine.

Challenges we ran into

The hardest product challenge was making the agent operational instead of conversational. A normal chatbot answer was not enough. RescueOps needed to preserve evidence, explain why the score was calculated, create an audit trail, and move the team into action without losing context.

We also had to balance demo reliability with live Slack platform behavior. RescueOps includes an RTS check path and graceful evidence-provider fallback, so the workflow remains testable while still supporting live Real-Time Search when Slack provides the required token context.

Accomplishments that we're proud of

  • Built a Slack-native workflow, not just a chatbot
  • Added explainable risk scoring with weighted evidence
  • Created an end-to-end action loop from scan to impact receipt
  • Preserved an audit trail inside Slack
  • Added dynamic support for multiple accounts and evidence providers
  • Added MCP-ready business context
  • Added optional grounded open-source reasoning for rescue communication
  • Kept the app lightweight, testable, and GitHub-ready

What we learned

The biggest insight was that Slack is not only where work is discussed. It can become the system where risk is detected, explained, assigned, and resolved.

We also learned that winning agent experiences need trust. RescueOps shows evidence, owners, actions, and measurable impact so teams can act with confidence.

What's next for RescueOps: Revenue Rescue Agent for Slack

Next, RescueOps can expand into production connectors for CRM, support, incident management, Jira, renewals, and customer success platforms. It can also add scheduled account scans, workspace-specific risk models, automated executive digests, and revenue protection dashboards.

Built With

  • ai-agent
  • block-kit
  • customer-success
  • mcp
  • mcp-server
  • python
  • real-time-search-api
  • revenue-operations
  • slack-agent
  • slack-ai
  • slack-api
  • slack-block-kit
  • slack-bolt
  • slack-socket-mode
  • slack-web-api
  • socket-mode
Share this project:

Updates