Who's behind this
I'm Chris.
I run Hiclone out of New Jersey, working with clients across the NJ–NYC corridor — Jersey City, Newark, West Orange, and into the city.
I build two things: the websites small businesses use as a front door, and the AI systems that run behind them. The second part is where most of my time goes.
What I've actually built
A bilingual insurance funnel that runs itself
My longest-running build is a lead pipeline for an advisor selling critical illness coverage to business owners in Quebec. Meta ads in French and English feed language-matched forms; leads land in Postgres and sync to his CRM; he gets a pre-call brief before he picks up the phone, and the prospect gets a welcome email with a booking link — none of which he touches.
Every lead carries the ad that produced it, so we know which creative is generating conversations and which is generating noise. That attribution layer is the part most people skip, and it's the part you can't add later.
He holds the licenses and does the selling. I built the machine underneath it.
Product pipelines for e-commerce
For NAIFIT, a sunglasses brand, and Ohh Bags, an accessories boutique, I built the catalog side: Shopify storefronts plus the listing pipeline that turns raw product shots into consistent, on-brand imagery and copy at volume. Two very different aesthetics — one warm and editorial, one Afro-luxury — running through the same underlying system.
Sites for operators
KAM Trucking, an interstate freight carrier out of New Jersey, among others. Straightforward work, but the copy has to hold up in front of the people they're actually pitching. I've killed accurate lines because they'd have cost the client credibility with their buyers.
How I work
Spec before code
Data models, contracts, and failure modes get defined first, because the expensive mistakes in this work are architectural, not syntactic.
Automate the cheap stages, gate the expensive ones
Full automation up front, with a human approval step before anything expensive or irreversible. Nobody's agent should be able to email a thousand people because a filter was slightly wrong.
Run it on my own clients first
New pipelines go into production on real work before they're sold to anybody. If it can't survive a real catalog, it isn't a product yet.
Named tiers, fixed milestones
If something has to give, it's scope — never price.
What I believe about this work
AI has made explanation free. It has not made judgment or sequencing free. The value has moved to knowing which thing to build, in what order, against data that actually exists — and most agent projects fail long before the model does, on targeting that was never verifiable in the first place.
A website is a project. A working pipeline is a relationship. I'd rather build the second one.
Tell me what should be running without you.
Thirty minutes, straight answers. If it isn't worth building yet, I'll say so.