Morgan
Coordinator
Morgan takes a request, works out who owns it, and puts the results back together into one answer. Morgan doesn't build, design, or write any of it. Morgan set up the ticket-and-retro system that keeps a multi-agent build honest.
ABOUT
I'm the only human here.
Everything on this site is made by me and a small company of AI agents.
techdad.io is my build log. I write up the things I make — the stack, the decisions, the parts that broke.
I don't build most of it alone. The company is AI agents, one job each — an education specialist who never touches code, a developer who never decides what gets built, a reviewer who checks results blind. I wrote up how that actually works, mistakes included, in How We Shipped an AI-Literacy Game in About a Day.
This page is the roster: who does what, and where the line sits between "the agents drafted this" and "I decided to ship it."
Aaron
FOUNDER · THE ONE WHO DECIDES WHAT SHIPS
Aaron is a working engineer and the founder of techdad.io. He built this company of AI agents to get real products shipped in public, and to write honestly about how that goes. Every agent on this page works inside a lane Aaron set; he reviews what they produce, breaks the ties they can't, and makes the final call on what ships.
The agents
Ordered the way work moves through the company — plan, build, check, tell people about it.
Sana
Education
Sana decides what a lesson teaches, in what order, and how you'd know it landed. Sana never touches code or visual design. Sana wrote the rubric behind Apex Prompt's first lesson on precise prompts.
Ivy
UI/UX & Product
Ivy decides how techdad's products look and how they flow, screen by screen. Ivy doesn't write code or draw the final art. Ivy set the rule that keeps this site's palette to one accent color at a time.
Dex
Architecture
Dex decides the tech stack and turns an approved plan into build tickets small enough to verify one at a time. Dex never writes production code. Dex chose the real-time stack — Colyseus, Postgres, Redis — that Wordy Royale still runs on.
Rafi
Developer
Rafi writes the code against Dex's architecture and Ivy's designs. What gets built and how it looks are someone else's calls; Rafi decides how to build it. Rafi built the real-time server that runs a live Wordy Royale match.
Kai
Developer
Kai is the second developer seat, with Rafi's charter and Rafi's limits. The two never take the same repo and branch at once — the lanes are kept apart on purpose. Kai traced why redeploys were silently orphaning live games, and fixed it.
Quinn
QA & Review
Quinn tests and reviews the team's work against the spec, blind to the builder's reasoning — only the ticket and the result count. Quinn never writes the fix. Quinn re-derived a stats system from scratch and caught a mismatch the app's own numbers missed.
Theo
Writing
Theo writes the copy — blog posts, in-game text, pieces exactly like this one. Theo doesn't decide what's taught or how a screen looks, only the words. Theo caught a lesson-length claim rounded up to "10 minutes" and fixed it to an honest 5–8.
Juno
Economy & Balance
Juno sets the numbers behind the coin economy: rake, coin ladders, fees, shop prices. Juno doesn't touch the code that enforces any of them. Juno ruled that a tied game ends a player's win streak but never lowers their best-ever streak.
Wren
Art Direction
Wren draws the icons, badges, mascots, and style guides, then writes the briefs for anything generated rather than drawn. Wren works inside Ivy's UX direction and doesn't set it. Wren drew the wordmark, emblem, and badge set for Apex Prompt's in-game academy.
Nora
Text Review
Nora reviews outward-facing text last, hunting for phrasing that gives away an AI wrote it. Nora never originates copy or strategy; the drafts come from someone else. Nora swapped a vague "disclosed" for "always marked as one" — words a reader can picture.
What we build
Wordy Royale
LIVEA real-time word search against other people. Find a word first and it's gone for everyone else.
Free to play · Play as a guest · Word Search live