Thanks! We provide eval templates that can be applied on specific stages or the whole conversation. Users can specify their own evals that can be as granular as they'd like. We're also working on conversation simulation feature that lets users quickly iterate on evals via simulating previous real conversations and seeing if the eval output aligns with human judgement.
P.S. Arkadiy is locked out of his HN account due to the anti-procrastination settings. HN team, can you plz help? :)
thanks for checking it out. i imagine 2 things, 1 - any edit actions require a "human in the loop" approval. 2 - for sensitive info, internal instances of llms or only local models accessing info
congrats on the launch! i build primarily with nextjs and vercel and will give it a shot. but we also just launched an on-call tool (i work at datadog), so i'll compare
Dash0 currently doesn't offer an own agent or a distribution of OpenTelemetry Collector.
You can use anything which is sending OTLP.
If your applications have been instrumented accordingly, you don't necessarily need a separate agent but can let the OpenTelemetry SDK send directly to Dash0.
I disagree. The final product (the food) is not better in any way for doing this. I do see the value of not making the server lie to the customer by pretending that they were going to get the "stunt chicken", but this whole problem seems to be caused by trying to make the preparation of the food into a show, and here is a case where the showmanship is getting in the way of efficiency.
If they just got rid of parading out the chicken in the middle of the process they could stage more things without implicitly lying to the customers, while improving timing. You would just lose some showmanship, which might be part of the brand.
