DevStandup AI

By analyzing GitHub activity and using Claude AI, it generates insightful summaries, PR reviews, generates unit test cases and docstrings, and posts updates on Slack.

Comment

Inspiration

  • Daily standups, PR reviews, and project coordination often take more time than actual coding. As teams scale, developers spend valuable hours syncing instead of shipping. Inspired by this challenge — and Lloyd Atkinson’s reflection on standups becoming a “parody of productivity”, we built DevStandup AI to make team collaboration effortless, insightful, and automated.

What it does

DevStandup AI automates the repetitive parts of software collaboration using GitHub activity and Claude AI. It:

  • Generates standup summaries directly from recent commits and pull requests
  • Provides AI-powered PR review insights and highlights
  • Auto-generates unit tests and docstrings to improve quality and consistency
  • Posts standup and progress updates to Slack for transparent team communication
  • Tracks overall project activity and contributions across repositories

How we built it

We built DevStandup AI using a fully serverless, event-driven architecture powered by Claude(AWS Bedrock), AWS and GitHub APIs.

  • GitHub API: to fetch commit activity, pull request data, and contributor events in real time.
  • AWS Lambda: to process incoming data and trigger automation workflows on demand, keeping costs low and scalability high.
  • AWS API Gateway: to expose secure endpoints for internal and external integrations, including Slack updates and the dashboard.
  • AWS Bedrock (Claude 3.5 Sonnet): as the AI engine for summarization, PR analysis, unit test and docstring generation, and natural-language standup reporting.
  • Slack API: to post automated standups, review insights, and activity summaries directly to team channels.

Challenges we ran into

  • Prompt optimization: Balancing concise, human-readable AI summaries with technical accuracy took multiple iterations of prompt design and few-shot examples.
  • Serverless orchestration: Coordinating multiple Lambda functions with API Gateway while managing concurrency took time to configure and get working smoothly.

Accomplishments that we're proud of

  • End-to-end automation of developer standups, PR reviews, unit test and docstring generation, and Slack updates.
  • Fully serverless deployment using AWS Lambda and API Gateway: scalable, cost-efficient, and easy to maintain.
  • Seamless AI integration with Claude 3.5 Sonnet on AWS Bedrock, delivering fast, context-aware summaries and insights.
  • Intelligent multi-source synthesis: Combined GitHub events, commits, and PR discussions into a unified team activity view.
  • Developer-first experience: Automated updates feel natural and helpful, reducing coordination fatigue while improving team transparency.

What we learned

  • Grounding AI in real data is key: Connecting Claude Sonnet 3.5 to actual GitHub activity dramatically improved the relevance and accuracy of AI-generated summaries.
  • Prompt engineering matters: Iterating on structured prompts and contextual payloads helped produce consistent outputs across standups, PR reviews, and documentation.
  • Serverless ≠ simple: While AWS Lambda and API Gateway simplified scaling, managing event triggers, timeouts, and API rate limits required thoughtful orchestration.
  • Human + AI collaboration works best: We feel AI is an assistant, not a replacement, developers trusted the AI more when they could review and refine its outputs, especially for tests and docstrings.
  • Integration design is as important as AI: Delivering insights where teams already work (Slack, GitHub) proved more impactful than building standalone dashboards.

What's next for DevStandup AI

  • Deeper GitHub integration: Adding issue and project board analysis for richer team insights.
  • Personalized developer insights: Tailored summaries, code review patterns, and productivity metrics for each team member.
  • Interactive Slack commands: Letting team members query AI directly in Slack for on-demand summaries or PR analysis.

Built With

Share this project:

Updates