Skip to content

marketing-bot
Status Active
Path AI/ML
Prereqs Pydantic, Instructor
Repo pysprings/lightning-talks (marketing-bot)

Overview

A clean-architecture AI planning system that applies the Deming Cycle (Plan-Do-Check-Adjust) to LLM-driven task management. The objective: "Increase participation in the Colorado Springs Python User Group."

Architecture

domain.py       → Domain models (Objective, State, Adjustments)
usecases.py     → Abstract use cases (Planner, Executor, DemingCycle)
adapters.py     → LLM integrations (SimplePlanner, SimpleExecutor)
driver.py       → Entry point with dependency injection

Key Concepts

  • Deming Cycle (PDCA) — Plan, Do, Check, Adjust applied to AI systems
  • Dependency injection — Clean separation between business logic and LLM adapters
  • Abstract base classes — Planner and Executor protocols
  • Structured output — Pydantic models via the Instructor library
  • Tool-based execution — Extensible action pattern
  • State management — Track LLM-driven progress across iterations

Try It

  1. Clone the repo and cd marketing-bot
  2. Install: pip install -e .
  3. Set API key: export OPENAI_API_KEY=sk-...
  4. Run: python driver.py

Where to Go Next

  • RLM — A different LLM orchestration architecture
  • DSPy Series — Systematic framework that formalizes many of these patterns