Theme 1 – Decision models are easy to build and copy
Many commenters stress that creating a Jev‑like model requires little novel work: you can repurpose existing LLMs, force structured output, or fine‑tune small models quickly.
- “You can use already trained large transformer models to make one, so it doesn’t require the kind of high‑scale compute … that an LLM does.” – didibus
- “It’s very easy if you have fairly basic ML knowledge.” – redox99
- “You can make a basic one in minutes based on existing open‑source models.” – XCSme
- “The concept existed a year before Jev … the underlying approach was already there.” – kerenskiy / petercooper
Theme 2 – Jev’s main innovation is productization, not core technology
Several users argue that Typesafe’s Jev succeeded mainly by polishing the API, ergonomics, and marketing around an idea that was already known.
- “Jev mostly innovated on the interface, API, and product concept around this, and made it click for a large number of people.” – woah
- “Jev created accessible/programmatic ergonomics around a general‑purpose classifier … intuitive api and structured data approach.” – ramoz
- “They were the first to bother to stop and pick it up, and market the shit out of it.” – TeMPOraL
Theme 3 – Rapid emergence of competitors and alternatives
The discussion notes that once Jev appeared, many similar models (Clef, At0m, Laya, etc.) quickly followed, often with better cost or latency, showing how easy it is to replicate the offering.
- “It’s very easy to copy an API, and any pretrained LLM can be adapted to work in this way.” – woah
- “Now everyone simply finetunes Qwen and makes a … vastly cheaper decision model.” – porridgeraisin
- “If CF’s benchmark is representative and sufficient, Clef outperforms Jev!” – tomrod
- “Clef‑flash is at $0.09 which is way more competitive.” – ssiddharth
These three themes capture the dominant sentiments in the thread.