Jev decision model filters LLM requests
The story so far
World-labs shared techniques for using the Jev decision model, which provides probabilistic answers through structured questions, reducing LLM response costs by 444.6 times and increasing speed by 193.6 times. PostHog used 2 yes/no questions to filter 98.7% of LLM calls in its Replay Vision product, with a threshold set to 0.4 to avoid data leakage risks. The decision model is suitable for single-judgment tasks, requiring data preprocessing or using LLM for complex reasoning.
AI-written from the reports · Updated 3 hours ago
Coverage timeline
Follow the coverage to see every side of the story.
- Hacker News · AI hot postsNot Everything Needs an LLM
The Jev decision model provides probabilistic answers by structuring problems, reducing LLM response costs by 444.6 times and increasing speed by 193.6 times. PostHog uses 2 yes/no questions to filter 98.7% of LLM calls in its Replay Vision product, with a threshold set to 0.4 to avoid data leakage risks. The decision model is suitable for single-judgment tasks, requiring data preprocessing or using LLM for complex reasoning.
Heat over time
- Now
- 3
- Peak
- 319 Nov, 19:00
- 24h change
- –
Hover or tap the chart to see each hour; use the arrow keys on a keyboard.