Skip to content
Phong DuongAI Product Leader

Open to senior AI product roles, consulting, and collaboration · Houston, Texas

I turn complex enterprise operations into practical AI products.

I spent fourteen years building products around complicated fuel, logistics, and operational workflows. Today I use that experience, along with agentic development, to build AI products end to end.

View my work Resume

Selected work

Products I’ve built

All work
Enterprise fuel operations + AI

Homebrew

An enterprise fuel operations platform built around the whole purchasing lifecycle, from the need through sourcing, dispatch, delivery, and reconciliation.

Explore Homebrew
Replenishment planning: regional groups ranked by replenishment status.
Agentic workspace + orchestration

Sophie

A workspace for doing real work with AI. Conversation, project context, memory, agents, tools, and oversight in one environment.

Explore Sophie
The Sophie workspace: the spine with the memory rail open.
Agent memory

Fraimed

Durable memory for agent work, so what was decided and what happened is still available later.

Explore Fraimed
Fraimed Outcomes: graded results per work item, which decide what memory stays trusted.
ChasingUpside.com

Screenshot placeholder

The view that best establishes what ChasingUpside.com is.

Product placeholder

ChasingUpside.com

Positioning line to be supplied. One or two sentences, in the same voice as the products above.

Explore ChasingUpside.com
BidPrep.AI

Screenshot placeholder

The view that best establishes what BidPrep.AI is.

Product placeholder

BidPrep.AI

Positioning line to be supplied. One or two sentences, in the same voice as the products above.

Explore BidPrep.AI

How I work

I don’t start by asking an agent to build.

I work an idea until I understand the problem, what already exists, and what is actually worth testing. Then I build a thin slice and use it to challenge the idea before committing to the larger product. Once I do, strong specifications, phased plans, persistent context, and clear acceptance criteria are what let the agents move quickly without losing the product along the way.

See how I work

About

I spent fourteen years turning complicated fuel and logistics operations into enterprise software. I now design AI products the same way.

The domain knowledge is the advantage rather than the constraint. I start with how people actually work, then decide where AI belongs.

About me Resume

Contact

Open to conversations about AI product leadership and enterprise automation.

Get in touch LinkedIn