Skip to main content

Alpha This is a new service. Help us improve it and give your feedback by email.

Minute AI Walkthrough

Step-by-step guide to transcribing meetings and generating AI-powered minutes with Minute AI

Explore Minute AI

20 minutes to see AI-powered meeting minutes

Built by i.AI (the UK Government's AI team) and open-sourced for public sector use. Upload a meeting recording, get a transcript, and generate structured minutes automatically.
Built by i.AI, open source, and ready to try

Minute is developed by the Incubator for AI (i.AI), the UK Government's central AI team. It uses AWS Transcribe for speech-to-text and Amazon Nova Pro via Amazon Bedrock to generate structured meeting minutes from transcripts. The tool includes multiple meeting templates, speaker identification, and a rich text editor for refining output.

What you'll do

In this 20 minutes walkthrough, you'll:

  1. Get your credentials -- Find your Minute AI URL and login credentials from CloudFormation (2 minutes)
  2. Log in to Minute AI -- Access the Minute AI interface and explore the dashboard (2 minutes)
  3. Upload a meeting recording -- Upload an audio file and choose a minute template (3 minutes)
  4. View the transcription -- See the AI-generated transcript with speaker identification (3 minutes)
  5. Review AI-generated minutes -- Explore the structured meeting minutes generated by Amazon Nova Pro (5 minutes)
  6. Edit and export -- Use the AI editor, download, and explore templates (5 minutes)

Before you start

Important You must have already requested the Minute AI scenario via NDX:Try.

Deploy the Minute AI scenario if you haven't already.

Important Deployment takes 25 to 35 minutes. This provisions an Aurora PostgreSQL database, ECS Fargate services, and CloudFront distribution.
Why does deployment take this long?

Minute AI provisions enterprise-grade infrastructure:

  • Aurora PostgreSQL database -- stores meetings, transcripts, and generated minutes
  • ECS Fargate services -- runs the web application and background workers
  • SQS queues -- manages transcription and minute generation jobs
  • CloudFront distribution -- provides secure HTTPS access
  • S3 bucket -- stores uploaded audio files

The wait is worth it -- you get a fully functional AI meeting minute tool to evaluate.

While you wait: Explore the CloudFormation Events tab to watch resources being created. You can also browse other areas of the AWS Console -- deployment continues in the background.

You'll need

  • Your CloudFormation stack to show CREATE_COMPLETE
  • The MinuteUrl from your stack's Outputs tab
  • The BasicAuthUsername and BasicAuthPassword from Outputs
  • A meeting recording in MP3, WAV, or WebM format (any short recording will work)

What you'll learn

For service managers

  • How AI can automate meeting minute production
  • Multiple meeting templates for different contexts
  • The time savings compared to manual minute-taking
  • How to edit and refine AI-generated output

For technical teams

  • How AWS Transcribe handles speech-to-text
  • How Amazon Bedrock generates structured minutes
  • ECS Fargate container architecture
  • SQS-based job queue processing

Key capabilities

Transcription

Automatic speech-to-text with speaker identification via AWS Transcribe

Minute Generation

Structured minutes with action items, decisions, and summaries via Amazon Nova Pro

Meeting Templates

Cabinet, Planning Committee, General, and custom templates

Start walkthrough

Takes approximately 20 minutes

Need help?

If you're stuck, check the troubleshooting section or contact us.