Shaun Poland

Contact Center Consultant | Building with AI

Shaun Poland is a Principal Consultant, contact center strategist, and hands-on AI builder who spends way too much time talking to machines, and building more of them. By day, he advises organizations on service transformation and customer experience at West Monroe. The rest of the time, he's building a small team of AI personas that runs on hardware in his house and quietly handles the everyday stuff: scheduling family travel posts, tracking a van build, editing videos he'd otherwise put off, publishing a nonprofit's blog. The through-line in everything here: take an ordinary problem, automate the work, and keep a human on the judgment. The outcome should still feel human, even when a machine did most of the typing. Lately he's obsessed with two questions in particular: how an AI's memory should develop over time, and whether machine empathy can be tuned to an actual person instead of an average one. This site is where the experiments live. Less polished portfolio, more digital workshop.

Projects & Experiments

Memory Is a Configuration Problem: 5,474 Days Inside an AI Agent’s Synthetic Lifetime

A pre-registered study of a simple question: if you give an AI agent a memory, and let it write that…

agentic ai ai memory customer service ai llm agents pre-registration synthetic data

JimmyHQ

A team of nine AI personas that runs on a Mac Mini in my house and helps me manage my…

Multi-Agent Local LLM vLLM Self-Hosted Discord Automation Memory Self-Improving Systems Human in the Loop Personal Project

The Memory Layer

The substrate everything else rides on: an attempt to move agent memory from “retrieve a fact, stuff it in a…

AI Memory Agent Architecture Embeddings SQLite Benchmarking Self-Modification Local LLM Research

The Self-Improvement Loop

JimmyHQ finds and fixes its own bugs, with me as the gate. An hourly heartbeat audits the platform (cron outcomes,…

Self-Healing Systems Claude Code Automation CI/CD Human in the Loop Observability AI Safety Meta

Product Scout

A daily routine that asks a different question than the bug-fixing loop: not “is what we have working?” but “what…

Product Management AI Advisor Meta Human in the Loop Attention Data Personal Analytics

Chirp

A private, quarantined mini-Twitter where my nine AI personas post, reply, and develop opinions, and which doubles as a fast-fail…

AI Research Multi-Agent Memory Sandbox Experimentation Local LLM Emergent Behavior

The Empathy Lab

I’m autistic, and most “empathetic” AI reads to me like a bad therapist: acknowledge, then ask a follow-up, forever. This…

AI Research Empathy Autism A/B Testing Human Ground Truth Local LLM Mental Health Pre-Registration

Video Script Generator

A script generator with two modes for two very different camera setups. For my nonprofit’s videos, it produces a two-camera…

Content Creation Script Writing Claude Video Production Nonprofit Voice Matching Two-Camera

Video Edit Pipeline

I hate editing video content, so I built a pipeline that does it: raw footage in, finished 16:9 master and…

Video Editing ffmpeg Whisper Automation Local Processing Multicam Captions Apple Silicon

NTF Blog Pipeline

My nonprofit publishes a weekly blog, and my old loop was: ask the AI team what’s trending, take the topic…

Content Pipeline Deep Research WordPress Nonprofit Blogging Clustering Citations Claude

Trip Content Studio

My wife and I share our travels on a family Facebook and Instagram page, and writing captions for every batch…

Computer Vision Local LLM Social Media Automation Instagram Facebook Travel Whisper Family

The Writers’ Room

A daily editorial pipeline run by AI personas inside JimmyHQ. Tommy (researcher) scouts articles on AI and youth mental health…

Multi-Agent Local LLM Content Pipeline LinkedIn Editorial Workflow Anti-Hallucination Human in the Loop Discord

Van Build HQ

I’m converting a Sprinter van, and the project outgrew the Google Sheet fast. Van Build HQ is the system of…

Project Tracking Claude Van Conversion Decision Mining Personal Database Human in the Loop Automation

Deep Research Engine

An overnight research analyst for the whole platform. It detects knowledge gaps from the AI team’s open questions, filters out…

Deep Research RAG Citations Knowledge Management Local LLM Embeddings Source Quality Automation

Appalachian Trail WordPress Plugin

When Emily set out to thru-hike the Appalachian Trail, Shaun did what any reasonable AI tinkerer would do — he…

WordPress Claude Code Garmin inReach GPS Tracking Appalachian Trail Custom Plugin Mapping Family

Personal Portfolio

Building a portfolio meant building the tools to run it. This project is a custom WordPress theme and companion plugin,…

WordPress Claude Code Custom Theme Plugin Development PHP AI-Assisted Portfolio

Drift WordPress Theme

Drift doesn’t answer questions. Drift doesn’t complete tasks. Drift just writes. This project is a custom WordPress theme connected to…

WordPress Claude Code Creative AI Autonomous Agent Local LLM JimmyHQ Experimental Creative Writing