ExecuAI

Inspiration

Meetings happen and decisions are made, but action items are constantly lost because no one sends follow-up emails. We wanted to completely eliminate this manual, error-prone process and ensure team accountability.

What it does

ExecuAI is an autonomous agent that ingests Meeting recordings, transcribes the audio, extracts specific action items with assigned owners and deadlines, and automatically sends targeted follow-up emails to the responsible team members.

How we built it

We built the pipeline using WunderGraph for API federation and Redis for caching. The system is designed to accept either a dummy text transcript or any raw audio file to generate the transcript. From there, we use Anthropic's Claude API to analyze the text, extract the required action items, and automatically complete the necessary follow-up actions.

Challenges we ran into

Fully utilizing the sponsor tools we set out to use. Eg.We planned on using GuildAI to execute the agentic actions from the meeting but due to the lack of API of GuildAI we had to pivot to use something else.

Accomplishments that we're proud of

We successfully built a fully autonomous end-to-end pipeline in just four hours.

What we learned

We learned how effective AI orchestration can be when chaining specialized tools together. Using a dedicated transcription model alongside a reasoning model proved incredibly efficient for transforming unstructured conversation into structured, executable data.