Where Things Stand
Professional Solutions Architect by day, casual project-er by night. Based in Portland, OR.
So, with the hardware set up, a few task-specific agents built, and an evaluation-tested orchestrator to tie things together, where do things stand? Well, the task-specific agents and the orchestrator work perfectly! Check out this CLI demo:
Unfortunately, the voice activation component isn’t quite as smooth. I can ask it a question and have it return a response, but the firmware on my Home Assistant Voice Preview Edition doesn’t handle streaming very well - so the process goes something like:
Me: “Hey - what’s the last email I got from Carol?”
…
…
<Grab a cup of coffee>
…
Clarvis: “Let me take a look! The last email you got from Carol was…”
Ok - it’s not that bad, but it’s a minimum 20 second delay, which doesn’t make for a very impressive demo. Trying to figure out how to cut down that latency will be my next big hurdle.
Taking a closer look at the project:
If your interested in seeing this project in more detail (or building a version yourself) check out my GitHub repo!