We’re using LangSmith for observability and honestly, it’s become way too critical to our AI stack.
Without it, we can’t debug user issues - it’s like trying to fix things blindfolded. Right now, even my local development is completely stuck because I can’t see any LLM traces.
This has me thinking… either we’ve made ourselves way too dependent on LangSmith, or maybe people just don’t realize how important it actually is?
We should probably start looking into alternatives before this becomes a bigger problem. We can’t have our whole workflow grind to a halt every time LangSmith has issues.
I’m Ankush, one of the co-founders. I’m really sorry about the degraded experience you and our other users are facing with LangSmith. I know how frustrating it can be to have developer workflows impacted.
Unfortunatelt, we are dealing with a complex production outage stemming from one of our vendors and is causing widespread degradation in the app.
We have tried to constrain the issue as much as possible and are moving towards a long term fix. We hope you decide to stick with us through your AI development journey. I am confident we will learn from this and improve.
Thanks for the response, Ankush.
I’m an early adopter of Langsmith, using it from the very beginning along with Langchain, and I love the platform.
I understand that outages can happen. We’ll wait for one more day before we start to look for alternatives.
I hope this issue is resolved soon. Also, it’d be interesting to know what caused this outage
Same here, we pay monthly for LangSmith and as mentioned this tool is too critical for us and we can not tolerate such a long downtime.
For a product you charge a solid amount of $$$ for, this is unacceptable