AI Agent & Tool-Connected Assistants
Designed AI-assisted support workflows combining intelligent intake, MCP-connected CRM context, instant specialist routing and human oversight for a high-volume enterprise team.
My role: AI Solution Designer & Implementation Lead
Project context: Enterprise sales-focused support operations; a 14-person team processing roughly 1,700–2,100 monthly submissions in recent months
What needed to change.
A 14-person support team faced high monthly request volume. Manual categorization, monitoring for new cases, missing request details and repetitive questions competed with higher-judgment specialist work.
Use an AI-assisted workflow and MCP-connected business context to speed up intake and answers, route work immediately, ask for missing information and preserve specialist control over complex or action-oriented requests.
What I designed and led.
- Designed guided Slack intake with AI-assisted question handling and dynamic prompts for missing request information.
- Integrated AI-agent assistance with existing knowledge and MCP-connected CRM record retrieval for contextual answers.
- Automated category and region routing, specialist assignment and instant notifications.
- Preserved human escalation for cases requiring judgment or changes to business records.
- Modeled time savings by mechanism rather than treating assumed AI deflection as observed closures.
What changed.
The workflow combined AI-assisted intake, contextual answers, immediate routing and specialist handoff. A scenario model estimates approximately 190–325 hours of monthly support capacity could be saved through routing (~70 hours), notifications (~40), missing-information prompts (~35) and AI answers (~45–180). These are estimated opportunities, not verified time savings or measured AI resolution. The broader AI-agent/MCP architecture also supports related triage and CRM-lookup solutions; their metrics should not be added together as independent gains.
~190–325 hours/month potential savings (modeled)