Critic calls out fake Jev demos, says it is a decision model not a generator
The author argues Jev is a decision model suited to tagging, classification, routing, browser and computer use, and tool selection, and that LLM plus Jev works better than replacing the LLM.
The amount of fake demos around Jev is killing me 😂 Half of them got caught larping because Jev is not your next text or code generation model It is a decision model. Give it a finite set of choices and it is extremely interesting for tagging, classification, filtering, routing, browser and computer-use actions, deciding retry/continue/stop (or) picking the right tool/subagent Where it gets interesting is for agents, Instead of dumping 100s of tools & skills into an LLM context and asking it to figure everything out, you use tool search/compaction to keep the context sane and let Jev narrow down the right tool/skill mixture. Then let an actual LLM do the reasoning, argument generation, code or text LLM + Jev makes far more sense to me than pretending Jev replaces the LLM