DAVID COOPER

Systems that remove work.

AI · Automation · Databases · Writing — built into systems people actually use.

Scroll

What I do

Four disciplines. One pipeline.

AI

AI that does useful work

Multi-agent systems and practical LLM integrations — local and cloud — selected on cost, quality, and fit. Not demos. Working systems.

Workflow

Automation that eliminates the repetitive

I connect the tools you already have so information moves without anyone re-keying it, and the workflow survives the person who built it.

Databases

The operational backbone

Student systems, finance intelligence, operational dashboards — I design the data layer everything else depends on.

Writing

Words that make it usable

Architecture documents, security reviews, specifications, educational material. A system nobody understands is a system nobody uses.

Selected builds

Real systems, in production

ClassWatcher

Problem
Teachers in a large online school were spending hours a week pulling the same student data out of Brightspace by hand.
Build
A teacher workflow platform: browser automation, a MariaDB backing store, messaging, and reporting in one place.
Result
Routine monitoring that used to take an evening now runs on its own — teachers read a morning report instead.

Hermes

Problem
One person can't be a research team, a writing room, and an ops desk at once.
Build
A self-hosted multi-agent AI platform: specialized workers for research, writing, and operations, running on local and cloud models chosen per task.
Result
An always-on staff of assistants that hands off finished drafts, not chat transcripts.

Household Finance Intelligence

Problem
Years of transactions spread across accounts, and no way to get answers out of them.
Build
A database-backed analysis system: SQL schema, import pipeline, categorization, and reporting dashboards.
Result
Every dollar categorized and queryable — decisions made from reports, not guesses.

Teacher Operations Dashboard

Problem
Student, parent, and workflow information scattered across disconnected systems.
Build
A unified dashboard pulling it into one operational view.
Result
One screen replaces five tabs.

Self-Hosted Infrastructure

Problem
Advice about systems is cheap. Running them is proof.
Build
A Proxmox cluster, TrueNAS storage, Docker and LXC services, monitoring, and local AI servers — designed, documented, and operated end to end.
Result
The lab where everything above gets proven before it reaches a client.

How I work

Method over magic

  1. Discover

    Understand the people, the process, and where the friction actually is — not where the org chart says it is.

  2. Map the system

    Trace the data, the tools, and the handoffs. Most problems are connections, not components.

  3. Build the smallest thing that works

    Ship a working slice early. Expand what earns its keep.

  4. Document it so it survives me

    Architecture notes, runbooks, and plain-language docs. You own the system, not a dependency on me.

Stack

Tools I actually run

Self-hosted to cloud — including local AI, for when your data shouldn't leave the building.

Writing

Thinking in plain language

Practical AI for organizations that don't need a moonshot

Most businesses don't need AGI. They need one reliable assistant doing one tedious job. Here's how to find that job.

Your problem isn't missing software — it's disconnected software

You already own the tools. What you're paying for in hours is the glue between them.

Local AI: when your data shouldn't leave the building

Self-hosted models are good enough for real work now. What that changes for schools, clinics, and anyone with sensitive data.

The database decision underneath every workflow problem

Automation built on a bad data layer just makes the wrong answer arrive faster.

About

Systems thinker before technologist

I'm David Cooper, a Canadian systems builder and consultant. I've owned a company, led a non-profit, and served on charity boards — and I've spent years explaining hard things clearly, first as an educator and now for clients. I work across the whole pipeline: infrastructure, databases, automation, AI, and the documentation that makes it stick.

The best systems are the ones that quietly remove unnecessary work, organize information, and help people make better decisions. That's what I build.

Let's remove some work.

Tell me where the friction is. Twenty minutes, no deck, no pitch.

Book a 20-minute systems conversation