Resum: API and SDK for live conversations

Enterprise API and SDK that connects a live conversation to named AI agents. A transcript while the session is open, then one file per agent when it ends.

Hi, I’m Williams. I built Resum.

Resum connects a live conversation to the AI agents you name.

You open a session and wait until the speech socket is ready. Then you send audio from a microphone, a file, or a live stream. While people are still talking, finalized turns append to a transcript. When the session ends, each named agent gets one file. notes and summary have a fixed shape. Any other name you pass is plain text.

Audio does not move until that socket is open. The same recording is saved as it is sent. Stopping the session still finishes the work: it flushes, then writes the agent files.

As an SDK, a session is five calls. Create it, wait until it is ready, pipe the audio in, listen while it is live, and end it.

https://resum.viroke.com

If you want to run it with your own conversations, write to williams@viroke.com. I’m here for questions about the flow.

Congrats on the launch, Williams! The "one file per named agent" idea is neat. We run a job search platform and keep thinking about interviews: a candidate_notes agent and an interviewer_summary agent on the same call would save HR a lot of rewriting. Does it handle two speakers switching languages (e.g. Armenian and English) within one session?

the end session step doing the final flush is a detail i'm glad u spelled out

partial transcripts are useful live but i'd want the final files clearly marked complete before another agent acts on them