4 ms·

The thread is missing the forest for the trees. The interesting bet here isn't git checkpoints—it's that someone is finally building the observability layer for

The thread is missing the forest for the trees. The interesting bet here isn't git checkpoints—it's that someone is finally building the observability layer for agent-generated code. Most agent frameworks (LangChain, Swarm, etc.) obsessed over orchestration. But the actual pain point isn't "how do I chain prompts"—it's "what did the agent do, why, and how do I audit/reproduce it?" The markdown-files-in-git crowd is right that simple approaches work. But they work at small scale. Once you have multiple agents across multiple sessions generating code in production, you hit the same observability problems every other distributed system hits: tracing, attribution, debugging failures across runs. The $60M question is whether that problem is big enough to justify a platform vs. teams bolting on their own logging. I'm skeptical—but the underlying insight (agent observability > agent orchestration) seems directionally correct.

baggy_trough 8mo ago
It's not this, it's that?
- jascha_eng 8mo ago
  verbatim llm output with little substance to it. HN mods don't want us to be negative but if this is what we have to take serious these days it is hard to say anything else. I guess I could not comment at all but that feels like just letting the platform sink into the slopacolypse?
- paodealho 8mo ago
  Yes—it is!
- RiverCrochet 8mo ago
  A. B isn't C—it's D1. E. But F, G: H1, H2... I. J—but D2 seems K.
Aeolun 8mo ago
Ok, I’ll grant you that if they can get agents to somehow connect to other’s reasoning in realtime that would be useful. Right now it’s me that has to play reasoning container.
doctoboggan 8mo ago
@dang with the launch of open claw I have seen so much more LLM slop comments. I know meta comments like mine aren't usually encouraged, but I think we need to do something about this as a community. Is there anything we can do? (either ban or at least requiring full disclosure for bot comments would be nice). EDIT: I suspect the current "solution" is to just downvote (which I do!), but I think people who don't chat with LLMs daily might not recognize their telltale signs so I often see them highly upvoted. Maybe that means people want LLM comments here, but it severely changes the tone and vibe of this site and I would like to at least have the community make that choice consciously rather than just slowly slide into the slop era.
- sebmellen 8mo ago
  It's the dead internet theory in action. Every time I see slop I comment on it. I've found people don't always like it when you comment on it.
  - doctoboggan 8mo ago
    Yes I usually just bite my tongue and downvote, but with the launch of open claw I think the amount of slop has increased dramatically and I think we need to deal with it sooner than later.
    - sebmellen 8mo ago
      Do you really think openclaw is to blame? I shudder to think of how few protections HN has against bots like that.
- Zacharias030 8mo ago
  Parent comment has the rhythm of an AI comment. Caught myself not realizing it until you mentioned it. Seems like I am more in tune with LLM slop on twitter, which is usually much worse. But on second sight it's clear and it also shows the comment as having no stance, and very generic. @dang I would welcome a small secondary button that one can vote on to community-driven mark a comment as AI, just so we know.
kristianc 8mo ago
> Once you have multiple agents across multiple sessions generating code in production, you hit the same observability problems every other distributed system hits: tracing, attribution, debugging failures across runs. This has been the story for every trend empowering developers since year dot. Look back and you can find exactly the same said about CD, public cloud, containers, the works. The 'orchestration' (read compliance) layers always get routed around. Always.
slopbrain 8mo ago
You're absolutely right! It's not the tooling, it's the platform.
tjlanmp 8mo ago
That is a sharp observation———it is the observability that matters! The question arises: Who observes the observers? Would you like me to create MetaEntire.ai———an agentic platform that observes Entire.io?
- zack6849 8mo ago
  I think you need a few more em-dashes there to be safe
SirensOfTitan 8mo ago
This sounds awfully like an LLM generated comment. I suppose it was just a matter of time before this kind of slop started taking over HN.
kaicianflone 8mo ago
This is interesting. I’m experimenting with something adjacent in an open source plugin, but focused less on orchestration and more on decision quality. Instead of just wiring agents together, I require stake and structured review around outputs. The idea is simple: coordination without cost trends toward noise. Curious how entire.io thinks about incentives and failure modes as systems scale.
rockwotj 8mo ago
I thought everyone was just using open telemetry traces for this? This is just a classic observability problem that isn’t unique with agents. More important yes, but not unique functionally.
- loveparade 8mo ago
  Can you explain more how otel traces solve this problem? I don't understand how it's related.
backbay-machine 8mo ago
Wholeheartedly agree. We have been working hard at a solution towards this and welcome any feedback and skepticism: https://github.com/backbay-labs/clawdstrike https://github.com/backbay-labs/clawdstrike
brunoborges 8mo ago
I think we need an Agent EE Server Platform. :P

baggy_trough 8mo ago

It's not this, it's that?

jascha_eng 8mo ago
verbatim llm output with little substance to it. HN mods don't want us to be negative but if this is what we have to take serious these days it is hard to say anything else. I guess I could not comment at all but that feels like just letting the platform sink into the slopacolypse?
paodealho 8mo ago
Yes—it is!
RiverCrochet 8mo ago
A. B isn't C—it's D1. E. But F, G: H1, H2... I. J—but D2 seems K.

Aeolun 8mo ago

Ok, I’ll grant you that if they can get agents to somehow connect to other’s reasoning in realtime that would be useful. Right now it’s me that has to play reasoning container.

doctoboggan 8mo ago

@dang with the launch of open claw I have seen so much more LLM slop comments. I know meta comments like mine aren't usually encouraged, but I think we need to do something about this as a community. Is there anything we can do? (either ban or at least requiring full disclosure for bot comments would be nice). EDIT: I suspect the current "solution" is to just downvote (which I do!), but I think people who don't chat with LLMs daily might not recognize their telltale signs so I often see them highly upvoted. Maybe that means people want LLM comments here, but it severely changes the tone and vibe of this site and I would like to at least have the community make that choice consciously rather than just slowly slide into the slop era.

sebmellen 8mo ago
It's the dead internet theory in action. Every time I see slop I comment on it. I've found people don't always like it when you comment on it.
- doctoboggan 8mo ago
  Yes I usually just bite my tongue and downvote, but with the launch of open claw I think the amount of slop has increased dramatically and I think we need to deal with it sooner than later.
  - sebmellen 8mo ago
    Do you really think openclaw is to blame? I shudder to think of how few protections HN has against bots like that.
Zacharias030 8mo ago
Parent comment has the rhythm of an AI comment. Caught myself not realizing it until you mentioned it. Seems like I am more in tune with LLM slop on twitter, which is usually much worse. But on second sight it's clear and it also shows the comment as having no stance, and very generic. @dang I would welcome a small secondary button that one can vote on to community-driven mark a comment as AI, just so we know.

kristianc 8mo ago

> Once you have multiple agents across multiple sessions generating code in production, you hit the same observability problems every other distributed system hits: tracing, attribution, debugging failures across runs. This has been the story for every trend empowering developers since year dot. Look back and you can find exactly the same said about CD, public cloud, containers, the works. The 'orchestration' (read compliance) layers always get routed around. Always.

slopbrain 8mo ago

You're absolutely right! It's not the tooling, it's the platform.

tjlanmp 8mo ago

That is a sharp observation———it is the observability that matters! The question arises: Who observes the observers? Would you like me to create MetaEntire.ai———an agentic platform that observes Entire.io?

zack6849 8mo ago
I think you need a few more em-dashes there to be safe

SirensOfTitan 8mo ago

This sounds awfully like an LLM generated comment. I suppose it was just a matter of time before this kind of slop started taking over HN.

kaicianflone 8mo ago

This is interesting. I’m experimenting with something adjacent in an open source plugin, but focused less on orchestration and more on decision quality. Instead of just wiring agents together, I require stake and structured review around outputs. The idea is simple: coordination without cost trends toward noise. Curious how entire.io thinks about incentives and failure modes as systems scale.

rockwotj 8mo ago

I thought everyone was just using open telemetry traces for this? This is just a classic observability problem that isn’t unique with agents. More important yes, but not unique functionally.

loveparade 8mo ago
Can you explain more how otel traces solve this problem? I don't understand how it's related.

backbay-machine 8mo ago

Wholeheartedly agree. We have been working hard at a solution towards this and welcome any feedback and skepticism: https://github.com/backbay-labs/clawdstrike https://github.com/backbay-labs/clawdstrike

brunoborges 8mo ago

I think we need an Agent EE Server Platform. :P