Search papers, labs, and topics across Lattice.
The paper introduces "compiled AI," a framework where LLMs generate executable code during a compilation phase, enabling deterministic workflow execution without further LLM calls. They present a system architecture and a four-stage generation-and-validation pipeline to convert probabilistic model outputs into production-ready code, emphasizing reliability and auditability for enterprise workflows. Evaluations on function-calling and document intelligence tasks demonstrate significant token amortization (up to 57x reduction) and competitive accuracy, alongside strong security against prompt injection and code vulnerabilities.
Forget runtime inference costs: "compiled AI" slashes token consumption by 57x while maintaining accuracy by pre-generating deterministic code from LLMs.
We study compiled AI, a paradigm in which large language models generate executable code artifacts during a compilation phase, after which workflows execute deterministically without further model invocation. This paradigm has antecedents in prior work on declarative pipeline optimization (DSPy) and hybrid neural-symbolic planning (LLM+P); our contribution is a systems-oriented study of its application to high-stakes enterprise workflows, with particular emphasis on healthcare settings where reliability and auditability are critical. By constraining generation to narrow business-logic functions embedded in validated templates, compiled AI trades runtime flexibility for predictability, auditability, cost efficiency, and reduced security exposure. We introduce (i) a system architecture for constrained LLM-based code generation, (ii) a four-stage generation-and-validation pipeline that converts probabilistic model output into production-ready code artifacts, and (iii) an evaluation framework measuring operational metrics including token amortization, determinism, reliability, security, and cost. We evaluate on two task types: function-calling (BFCL, n=400) and document intelligence (DocILE, n=5,680 invoices). On function-calling, compiled AI achieves 96% task completion with zero execution tokens, breaking even with runtime inference at approximately 17 transactions and reducing token consumption by 57x at 1,000 transactions. On document intelligence, our Code Factory variant matches Direct LLM on key field extraction (KILE: 80.0%) while achieving the highest line item recognition accuracy (LIR: 80.4%). Security evaluation across 135 test cases demonstrates 96.7% accuracy on prompt injection detection and 87.5% on static code safety analysis with zero false positives.
