Skip to main content

Building ConsultFlow - The Journey

· 3 min read
Sudeep SRS
Sudeep SRS
Creator of ConsultFlow

ConsultFlow started as a solution to a common problem: the tedious, repetitive work that follows every meeting in consulting environments.

The Problem​

As consultants, we spend countless hours on post-meeting tasks:

  • Writing meeting minutes
  • Sending follow-up emails
  • Scheduling next meetings
  • Updating team channels

These tasks are essential but repetitive. What if we could automate them?

The Vision​

I envisioned a platform that could:

  1. Transform meeting transcripts into structured documents using AI
  2. Automate Microsoft 365 actions like email, calendar, and Teams
  3. Provide a visual workflow builder for non-technical users
  4. Execute workflows with one click

Technical Journey​

Choosing the Stack​

Frontend: I chose Next.js 14 with the App Router for its modern patterns and excellent developer experience. Tailwind CSS handles styling with a custom dark theme inspired by modern tools like n8n and Linear.

Backend: Express.js with TypeScript provides a flexible, type-safe API layer. MongoDB stores workflows and execution logs as documents—a natural fit for the flexible schema needs.

AI: OpenAI's GPT models power the MOM generation. The structured prompting ensures consistent, professional output.

Authentication: Azure AD OAuth 2.0 enables seamless Microsoft integration. Users sign in once and access all Microsoft 365 services.

The Workflow Builder​

The visual workflow builder uses React Flow, a powerful library for node-based editors. Users can:

  • Drag actions from a sidebar
  • Connect nodes to create flows
  • Configure each node with specific settings
  • Save and execute with one click

AI-Powered MOM Generation​

The MOM generator uses a carefully crafted prompt that instructs GPT to:

  • Extract key discussion points by speaker
  • Identify action items with assignees and dates
  • Summarize in a concise abstract
  • Format everything in professional Markdown

Microsoft Graph Integration​

The trickiest part was integrating with Microsoft Graph API:

  • Token management (access + refresh tokens)
  • Permission scopes (Mail.Send, Calendars.ReadWrite, etc.)
  • Error handling for various failure modes

Lessons Learned​

  1. Start with the user flow - I designed the UI first, then built the backend to support it.

  2. AI needs structure - Detailed prompts with specific formatting instructions produce consistent output.

  3. OAuth is complex - Token refresh, scope management, and error handling require careful attention.

  4. Dark themes matter - Users spend hours in these tools; a well-designed dark theme reduces eye strain.

What's Next​

ConsultFlow is just the beginning. Planned features include:

  • Scheduled workflow triggers
  • PowerPoint generation
  • More AI-powered actions
  • Team collaboration features

Try It Yourself​

ConsultFlow is open source! Check out the documentation to get started.

Built with ❤️ by @srs-sudeep