Skip to content
Mitul Shah

Case Study — Featured Work

Agentic Partner Management Platform

An AI agent platform on MCP that automates the operational work of managing strategic partner relationships — built solo, then scaled org-wide to 80%+ adoption.

  • 5–8hrs

    saved per SA, per week

  • 7systems

    connected over MCP

  • 80%+

    org-wide adoption

  • 10+

    contributors via code review

The problem

The job around the job: partner-facing Solutions Architects lose hours every week to meeting prep scattered across seven systems, note-taking, action-item tracking, and executive updates — time that should go to the advising that actually earns partner trust.

What it is

An AI agent platform: a Model Context Protocol (MCP) framework connecting seven enterprise systems — email, Slack, ticketing, CRM, wiki, and developer docs — with canonical workflows (20 agent skills, 24 automation hooks) and per-harness adapters, so the same workflows run in any of the four AI coding agents an SA already uses.

Capabilities

  • Meeting prep in seconds

    Auto-pulls emails, Slack threads, tickets, and meeting history — no tab-hopping.

  • Automated meeting notes

    Paste raw notes → formatted files with action items extracted and TODO lists updated.

  • Partner health checks

    Scored dashboards across five categories — Technical, Delivery, Communication, Action Items, Risk — with staleness detection and trend tracking.

  • Executive summaries

    Polished HTML dashboards ready to send to leadership, generated on demand.

  • COE document builder

    Correction-of-error documents with interactive 5 Whys analysis and self-review.

  • Partner handoffs

    Handoff documents that adapt for temporary coverage vs. permanent transfer.

  • Ticket tracking

    Paste a ticket URL → auto-tracked with status updates; bulk one-click refresh with stale-ticket flagging.

  • Proactive date alerts

    Upcoming deadlines and overdue items surfaced across all partners.

  • Papercuts mechanism

    SAs log operational friction to a shared doc from any conversation in under 60 seconds — giving managers visibility into systemic issues.

  • Self-learning loop

    End-of-conversation hooks prompt saving new knowledge back to the knowledge base; in-conversation awareness catches gaps in real time.

  • Privacy controls

    Per-partner choice of shared or local-only data with a single agent command.

  • Platform troubleshooting

    The agent reads platform docs and crash/log output via MCP to diagnose partner app issues and draft technical responses with code examples.

Adoption & community

Built solo — then turned into a team platform.

  • Hands-on workshops across the US, EU, and Japan — including a 1-hour SA Summit session with two live exercises — plus weekly office hours, a knowledge hub, and self-paced labs.
  • A contribution pipeline built for non-developers: an automated pre-review agent handles mechanical checks, and a 5-SA code-review pool reviews every change — 10+ contributors have landed changes.
  • Five releases shipped through a structured mainline/staging release process; kanban roadmap with 20+ tracked tasks.
  • 12 weeks of Monday "power-user tips" covering context management, workflow patterns, and agent optimization.
  • Wiki, README, workshop guide, and self-paced exercises for ongoing onboarding; weekly office hours.
  • Teammates actively contributing hooks — an email digest, a gamified stats tracker — through a maintained review process.

Outcomes

  • Institutional knowledge compounds — every interaction feeds the partner knowledge base.

  • A governed data contract requires human validation before partner data promotes to shared systems — structured records to the CRM, narrative context to a knowledge base — keeping agent output auditable.

  • Staleness detection catches dropped items before they become problems.

  • In production use across multiple strategic partner relationships.

Want the full story?