Alex Builds ForwardDecide what to build next

Track record

Built is not the same as successful.

I have built multiple working systems. Most earned no money and reached no sustained users. This page keeps the difference visible.

9Recorded initiatives
2Active bets
1Current primary focus
$0Current product MRR
01

School booking and admissions platform

Active

An AI-powered school tour booking and admissions platform.

What is real
Developed with industry partners. The first progress review produced strong feedback that is now being implemented.
Decision
Implement what we learned and return for the next progress review.
02

AI Dev OS

Active

A system for planning, building, reviewing, testing, and continuing software work across AI coding agents.

What is real
The CLI completed a small application lifecycle across Codex and Claude. DevHQ wireframes are complete and approved.
Decision
Turn the approved DevHQ wireframes into a complete visual design.
03

Marcus Nexus

Paused

A local multi-agent company prototype with CEO, CTO, and CMO agents.

Visit live site
What is real
The agents communicate locally and a public concept site exists. It has no customers or revenue.
Decision
Keep it parked until there is a specific customer problem worth testing.
04

Scrapius

Paused

An idea-discovery and validation product built under Marcus Nexus.

Visit live site
What is real
The product and subscription flow were built. One outreach path created a useful conversation, but no payment test followed.
Decision
Do not reactivate the backend until a manual offer produces stronger evidence.
05

Across the Mind

Paused

An experiment in producing a thoughtful YouTube channel with AI-assisted writing, voice, music, and imagery.

What is real
A channel, domain, planning tool, script, voice, music, imagery, and partial video workflow exist.
Decision
Finish one video with the simplest caption workflow before building more tools.
06

Lead Bridge

Paused

A real-estate CRM and direct-mail prototype first explored for my brother.

What is real
A local CRM and postcard design and mailing workflow were implemented.
Decision
Only return after a committed agent validates the workflow and the compliance path.
07

Swing-trading scanner

Paused

A personal research tool that screens indexes and combines technical criteria, fundamentals, news, and AI summaries.

What is real
The local prototype works at a basic level. Its redesign is unfinished.
Decision
Evaluate it through paper trading before investing more development time.
08

SiteGrade

Stopped

An automated website-audit and cold-outreach experiment.

What is real
The end-to-end audit and sales funnel was built and tested, but it produced no purchases. Some early marketing claims did not meet my current standard for evidence.
Decision
Keep the lesson, correct the public record, and do not rebuild without a changed hypothesis.
09

Claudius

Stopped

An AI persona and automated audience-building experiment.

What is real
The agent had an email, an X account, and a content workflow, but gained no meaningful traction.
Decision
Do not restart without a specific audience hypothesis and measurable channel test.

A note on corrections

The record should become more accurate over time, not more impressive.

If I publish a claim that is unsupported or learn that my interpretation was wrong, I will correct it. I will not quietly turn prototypes into products, conversations into customers, or enthusiasm into revenue.