Samuel Alejandro

One developer, a small company of AI agents

This is how I run my projects, and the setup I study as an applied AI researcher. Each project has its own AI specialists, a project manager checks their work before it reaches me, and I set the direction and decide what ships.

Who answers to whom

Roles, not names, as of November 2026. Every agent has one job, its own folder, notes and rules.

MeSets the direction and decides what ships
Project managerPlans the work, hands it out, checks it

Glassline

  • Senior developer
  • Developer
  • Helper agents
  • Hardening analyst
  • PR and announcements
  • X poster
  • Reel scriptwriter
  • 3D artist
  • Sound and music

GamesAndChill

  • Site developer
  • SEO
  • Security
  • Security research
  • Android developer

Bots and server

  • Bots and game data
  • Server admin

Everything Since

  • Developer

Research and support

  • Research and game design
  • Transcription and translation

Client work

  • E-shop developer
  • Video-tooling developer

How it runs

  1. One job each, with its own folder, notes and rules.
  2. Two models check each other: the senior developer runs on a different model from the developer whose work it reviews.
  3. A second look before mine: every reel take and cut is reviewed, and the project manager checks the work before it reaches me.
  4. Every job is logged with an estimate and the real time.
  5. Decisions come to me as options with a recommended pick.

For your project

Several parts of your project move at once, every piece gets a second look before it reaches me, and you deal with one person who answers for all of it.

Have a role or a project in mind?

Send me a few lines about the job or the thing you want built. I read every message myself and answer by email.