MeetFlow

• 

  Landing page

• 

  Signup with stych

• 

• 

  Home page

• 

  Analytics

Inspiration

Meetings are where decisions are made and tasks are assigned, but too often the details get lost in long transcripts, messy notes, or forgotten follow-ups. We wanted to create a tool that doesn’t just transcribe meetings but also summarizes discussions, extracts action items, and transforms them into real project workflows.

That’s how Meetflow was born — to bridge the gap between talk and action.

What it does

Meetflow takes meeting recordings and automatically:

Transcribes them into clean text using Gladia’s MeetingBaaS.

Summarizes key discussions and decisions.

Extracts action items with assignees and due dates.

Pushes tasks directly into Jira, turning talk into real workflows.

Provides a dashboard to visualize summaries, tasks, sentiment, and quarterly progress.

How we built it

Gladia for transcription and meeting summaries

LlamaIndex + OpenAI to analyze transcripts, extract tasks, and generate structured insights

Stytch for user authentication and security

Lovable to design and generate the frontend dashboard

Challenges we ran into

One of the biggest challenges was extracting accurate action items from long and sometimes messy transcripts, where context and intent weren’t always clear. We also had to tackle speaker attribution, ensuring tasks and decisions were assigned to the correct people in the meeting. Another recurring difficulty was getting LLMs to output valid JSON consistently, which required careful prompt engineering and retries. Finally, we faced the pressure of building a polished demo within tight time limits, which pushed us to prioritize the most impactful features.

Accomplishments that we're proud of

Despite the challenges, we managed to build a working end-to-end flow that goes from meeting audio to transcripts, then to structured insights, and finally to a functional dashboard. We also designed a simple but powerful dashboard that makes it easy for teams to review what was discussed and decided. Another accomplishment was successfully integrating Stytch authentication, which ensured security for our prototype. Most importantly, we’re proud of learning how to orchestrate multiple AI tools into one cohesive product.

What we learned

Through this project, we learned how to effectively use LLMs with LlamaIndex for transcript analysis, which made querying and structuring meeting data much easier. We discovered the importance of prompt design in achieving consistent and reliable outputs from large language models. We also realized how AI can transform ordinary meeting transcripts into real productivity tools that teams can actually act on. Lastly, we gained valuable experience in rapid prototyping with Lovable, which helped us spin up a functional frontend quickly.

What's next for MeetFlow

Looking ahead, we plan to integrate Meetflow with popular project management tools like Jira, Notion, and Trello so that extracted action items can flow directly into existing team workflows. We also want to add advanced analytics, including participation trends, decision recurrence tracking, and sentiment analysis across meetings. Another exciting step is enabling live integrations with Zoom, Teams, and Google Meet, allowing Meetflow to act as a real-time assistant during meetings rather than just post-processing recordings. Finally, we aim to improve scalability and security, ensuring that Meetflow can grow into an enterprise-ready solution that organizations can rely on every day.

Built With

• gladia.io
• llamaindex
• node.js
• react
• stych
• typescript