this harness stack is f*cking gold TypeSafe AI + LangChain finally dropped the layer that makes agents affordable at company scale: jev running inside the langchain agent loop as middleware the problem every big team hits: each fork in the loop is another full LLM call. at 10 agents... that's a bill. at 10,000 it's the budget > routing before: every task goes to the frontier model now: jev picks the least costly model that can finish it. lookups go fast, architecture goes powerful > guardrails before: the "is this command dangerous?" check sat in the closed parts of claude code, codex, cursor now: AutoModeMiddleware(tools=["bash"]). one line, any agent, blocked before the tool runs > decisions before: one LLM call per question now: every question about the same state in one request, answered in parallel > audit before: "the model decided" now: the answer, the probability and the confidence stay in agent state > adoption no rewrite. pip install langchain-typesafe and drop it into the agent you already run TypeSafe reports up to 200x faster inference and 400x lower cost than comparable LLMs on classification. the LLM still reasons and writes. jev takes the thousands of small calls in between already live: browserbase agents for fractions of a cent, a live trading agent, email triage at scale full loop mapped below: 7 agents, 4 jev gates per step count the forks in your agent loop tonighshow more

@N01ennn one thing I've noticed is the cost runaway when not controlling those loops tightly. Hit that wall building my own thing. Keeping an eye on loop depth saves sanity.

pure builder gold, trying this stack tonight broski

thx bro , im try , its was not hard make it because JEV cooking

LangChain's middleware finally exposes token burn per tool call. Still can't see which MCP server lied to your agent though.

fr good visualisation

The real unlock isn't another layer, it's killing redundant LLM calls before they happen.

Putting Jev inside the agent loop makes sense

Excellent. C'est le meilleur cas d'usage que j'ai pu constater jusqu'à maintenant. Merci

This actually solves the LLM cost problem for enterprise-scale agents, pretty neat.

Company-scale agent cost usually dies in the loop, not the model bill. I'd meter retries + tool calls per accepted artifact, and fail the run when the harness can't show that ratio.

Every fork billed as a full model call is how spend multiplies. Keep cheap decisions off that path, and hard-cap the loop either way.

this dashboard making my head hurt

crazy idea bros

middleware at the right layer serious cost control

This is like putting a fast tollbooth before the expensive road: cheap checks handle routine turns, while the frontier model takes the genuinely hard ones.

4 gates at 0.95 each = 0.81. enjoy

Absolute goldmine of a stack breakdown. Definitely testing this out!

This is realy focking gold

4 jev gates per step adds up fast

saving this stack bro

yeah counting the forks. that's where the cheap-model pick actually pays