An AI app for checking suspicious messages.

I built Maria to help older adults understand a suspicious message and decide what to do next. Someone can share a message with the app and get a plain-language second opinion.

I worked across the product, interface and AI workflow, then built the website and launch film to explain it. The app is built with React Native, Expo & TypeScript.

An agent tool for checking a payment before it happens.

An agent can prepare a payment without checking whether it matches the user’s intent and limits. I built a consult step that explains what needs attention before the agent pays.

Hedwig compares the request with defined limits and stops when it cannot check it. The agent receives reasons it can use to ask a better question. Signing and sending stay with the wallet.

The open-source MCP tool and plugin have a public release and install guide.

Portfolio tools an AI agent can explain.

Wallet data takes work to turn into a useful view of holdings and profit or loss. I gave an agent a small set of tools to explain that data and prepare a proposed next action for review.

Scout keeps transaction preparation separate from execution. A no-key sample tour lets someone try the local MCP package before connecting a provider.

Recurring purchases stop at a proposal.

The portfolio site you’re on right now.

I designed & built gabeonchain.com to hold both sides of my work, marketing & engineering, in 1 place. It runs on plain HTML, CSS & JavaScript with no framework, set in 1 typeface.

The light field is Canvas 2D and repaints in each project’s color, 400ms in and 600ms out. At idle it paints 0 frames, v quiet until you pick a project.

My moat is knowing where great ideas come from, when to kill them, and how to build as what’s possible changes every week.

I do my best work between clients, creative teams, customers, and engineers. I find the need, turn it into clear requirements and working systems, then explain the technical choices in business terms.

I use data to find the problem, then design a system people can understand and use. Each project ends with a clear handoff for whoever takes it next.

15 years in marketing taught me to work with customers and creative teams and to explain why a product matters. About 6 of those years were with Web3 teams.

I bring that experience to applied AI software engineering. I can work through the business problem with a team and build a fix they can use.

Open source · Local workflow

Each release starts with a walkthrough, a feature outline, and raw footage.

Workflow demo

Capture the contact and company once.

Experimental · Offline preview tested

Collect employer listings and remove duplicates.

Open source · Local research workflow

Research accounts through your chosen sources.