About

I'm Wensheng Chen; I go by Wilson. I'm CTO at The Juicy Crab, and I build things across AI, developer tools, and practical software systems. This page is a snapshot of what drives me, what I care about, and how I think about the work.

Philosophy

I believe the best tools disappear. The software worth building doesn't demand attention; it removes friction so quietly that people forget it's there. I default to static delivery, build-time validation, and only a small explicit dynamic seam where the product truly needs it.

I also believe ambiguity is the real tax in AI-assisted work. Before I build, I want the idea challenged, the requirement made concrete, and the constraints written down. The more clearly intent is encoded, the less generic the output becomes.

Writing is how I stress-test ideas. Shipping is how I stress-test intuition. The source of truth should stay close to the work itself: the codebase, the reference files, and the operating rules that agents and humans can both follow.

Interests & Expertise

The problems I gravitate toward sit at the seam between how people think and how software works. A few areas I keep returning to:

Context-Aware Agents

I build AI systems that understand the full picture — intent, history, and constraints — and encode voice and procedural knowledge so agents act with judgment instead of plausible defaults.

Spec-First Workflows

I care about the work that happens before implementation: pressure-testing the idea, sharpening the requirement, and turning that thinking into reference files and skills an agent can actually use.

Build-Time Guardrails

I like pipelines that fail early and clearly. Schema validation, metadata checks, and deterministic content workflows keep the build honest and make shipping feel less like guesswork.

Static-First Architecture

I favor systems that are fast by default, predictable to deploy, and resilient to failure. Default to static; add a server, edge function, or runtime only when there is a concrete reason.

Background

I started my career at the Financial Times, where I spent close to five years as an engineer working across their digital platform. That experience gave me a deep appreciation for systems that serve millions of readers without getting in the way.

From there I founded Innovative Web Services and served as CTO for about three years, building custom web solutions for businesses that needed more than off-the-shelf templates could offer. After that I moved through a series of engineering roles at Zume, then as a senior engineer at HubSpot and FullStory, each sharpening a different part of how I think about developer experience, data pipelines, and product quality.

I am still CTO at The Juicy Crab, where I lead technology strategy for a growing restaurant group. In parallel I run bestpos.io, a restaurant marketing agency, and kloudeats.com, a first-party online ordering platform, while continuing to build AI-powered developer tools on the side.

Principles I Build By

Pressure-test the idea
A fast build starts before the repo exists. Challenge the wedge, the user, and the failure modes before asking an agent to implement anything.
Encode the constraints
Voice guides, schemas, and skills turn taste into operating instructions so agents do less guessing and the output stays closer to intent.
Fail early
Validation belongs in the pipeline. Catch broken assumptions at build time, not after a deploy when the ambiguity has already spread.

Interested in discussing this further?

I'm always open to connecting with fellow builders and founders.

Start a conversation