ALAdam Langinadam-lang.hashnode.dev·Sep 21 · 10 min readJev as the Turn Interpreter: A Conversational Agent Without Tool CallingReplacing the tool-calling LLM with a jev classifier cut total cost from $0.2866 to $0.0265 and p90 turn latency from 6,421ms to 3,508ms on the same 11 scripted conversations. Both agents passed 11 of00
ALAdam Langinadam-lang.hashnode.dev·Sep 18 · 17 min readEvals as Unit Tests: A Practical Approach for Agentic SystemsThis article proposes a simple discipline for evaluating agentic LLM systems: treat evals as unit tests. They live in the repository, they run before merge, and a failure blocks the build, exactly as 00
ALAdam Langinadam-lang.hashnode.dev·Sep 18 · 23 min readHow to Improve Agentic APIs: Expose a Simple Workflow APIThe most reliable architecture is the one that asks least of the model. Summary Conversational agents are often built by writing a workflow into the system prompt and letting the model work out which 01M
