Research‑Backed, Agent‑Coordinated Health Plans
I built LifeLogic to generate and validate daily plans using a system of three agents. It’s now archived, but here’s how it worked—and what it taught me.
The Three Agents
Purpose‑built agents collaborated to produce plans that were both evidence‑based and practically accurate.
Research Agent
Indexed ~250M papers via SciSpace to surface relevant, trustworthy insights—fast. It grounded recommendations in current literature.
Calorie Calculator Agent
Validated an LLM‑generated plan by computing exact calories/macros using a nutrition API—turning ideas into precise, trackable numbers.
Master Orchestrator
Coordinated both agents, reconciled outputs, and produced a single, finalized plan—research‑backed and numerically validated.
From Evidence → Plan → Exact Intake
The Research Agent scanned SciSpace’s corpus to gather relevant findings and constraints.
The Master LLM drafted a plan; the Calorie Agent validated each meal against a nutrition API.
I recorded exactly what I ate each day using LifeLogic, enabling trends and feedback loops.
Gut Score Trend (archived)
Read‑onlyWhy it’s locked
I built this to solve a personal health problem—and it worked. The live submission features are now disabled. This site remains as a transparent, high‑level walkthrough of the approach.