Tag
The author built an AI trading agent but found that using LLM for direct trading decisions is unstable, leading to a deterministic system where LLM translates strategies into explicit rules for execution.
This paper proposes a Program-Solve interface where clinical language models generate Python code for a deterministic executor to perform math calculations, evaluating on MedCalc-Bench and finding improved accuracy for larger models like Qwen2.5-32B compared to direct arithmetic and hand-written libraries.
Kent C. Dodds explains his Cody platform for running deterministic AI agents, emphasizing the role of software primitives in building reliable and secure systems.
The article discusses the challenge of separating probabilistic language understanding from deterministic execution in voice agents, recommending a confidence check for conditional phrases before triggering structured actions, as implemented with Vomo AI.
The blog argues that as AI generates code faster, understanding the combined execution becomes difficult, and proposes using graph engineering with a compiler to create deterministic orchestrators.
This article proposes a protocol for AI agents that replaces knowledge-based inference with presence-based verification using a declarative Checklist, ensuring that agents only ask for missing information and execute deterministically based on declared requirements.
Botcircuits Argus pre-compiles repetitive workflows into deterministic execution flows, cutting ~80% token usage while maintaining accuracy by providing the agent only the exact context needed at each step.
Describes a technique to reduce LLM costs in browser agent tasks by using a single planning call followed by deterministic execution, achieving 50x cost reduction compared to standard agent loops.