Some of it can be busywork, but for me the intermediate artifacts (plans, design docs, etc) serve a real purpose: they create a verification surface where you can check that the agent is creating the right thing before it goes all the way. It's exactly the same reason we created short sprints: if the team misunderstood the requirements and built the wrong thing, you only lost a sprint. We lost months of work when we did waterfall because the product did not match what the customer had in mind.
I have deterministic and stochastic tests that run on each artifact. For those that have a high risk of "not the right thing", I manually review the artifacts. But if it's bog standard I just rely on the auto-gates to reject and get the agent to retry the artifact.
This gets me a high-volume pipeline that yes uses a lot of tokens, but at the same time doesn't overwhelm me. I only deal with things that genuinely need my attention. That's worth it for me, and not busywork.
The current rate at which AI is involved in our work is not sustainable. This is the Golden Era of AI in which they're making us adopt the technology, only to be then dependent on it. The advantages are clear, the possibilities are there but they're not endless especially when it all goes crashing into the budget wall.
Absolutely agree. A lot of LLM-driven work is just inflated busywork with little real output. High token usage doesn’t equal genuine productivity, just unnecessary repetitive verification and paperwork.
Token spend has no per-output budget gate while human review still
does. Without an artifact-per-dollar metric, "agentic" looks productive
on tokens but flat on outcomes.
Yes i agree with you many times i feel that time make more work than before and spending more budget. I don't know if is the correct place to say that that is an extension called the Ceres copilot for vscode that wont make loops and don't burns apis, this is what i use now with deep seek connected.. put when there is a complicated job am still stuck with codex...
Some of it can be busywork, but for me the intermediate artifacts (plans, design docs, etc) serve a real purpose: they create a verification surface where you can check that the agent is creating the right thing before it goes all the way. It's exactly the same reason we created short sprints: if the team misunderstood the requirements and built the wrong thing, you only lost a sprint. We lost months of work when we did waterfall because the product did not match what the customer had in mind.
I have deterministic and stochastic tests that run on each artifact. For those that have a high risk of "not the right thing", I manually review the artifacts. But if it's bog standard I just rely on the auto-gates to reject and get the agent to retry the artifact.
This gets me a high-volume pipeline that yes uses a lot of tokens, but at the same time doesn't overwhelm me. I only deal with things that genuinely need my attention. That's worth it for me, and not busywork.
The current rate at which AI is involved in our work is not sustainable. This is the Golden Era of AI in which they're making us adopt the technology, only to be then dependent on it. The advantages are clear, the possibilities are there but they're not endless especially when it all goes crashing into the budget wall.
That's true if your org treats artifacts as progress.
When artifacts are cheap, differentiation comes from quality.
How are you measuring progress at your company?
Do you feel AI agents are helping you achieve company goals faster now?
Absolutely agree. A lot of LLM-driven work is just inflated busywork with little real output. High token usage doesn’t equal genuine productivity, just unnecessary repetitive verification and paperwork.
Token spend has no per-output budget gate while human review still does. Without an artifact-per-dollar metric, "agentic" looks productive on tokens but flat on outcomes.
Yes i agree with you many times i feel that time make more work than before and spending more budget. I don't know if is the correct place to say that that is an extension called the Ceres copilot for vscode that wont make loops and don't burns apis, this is what i use now with deep seek connected.. put when there is a complicated job am still stuck with codex...
[flagged]
[flagged]