Jayson AI Holland
For plumbers, electricians & trades businesses · NZ/AU

You’re already paying for leads your business isn’t answering fast enough.

Every enquiry that sits unanswered after hours, mid-job, or over a weekend is a job going to whoever calls back first. I build a simple system that catches it, replies immediately, and follows up — so it doesn’t just sit in an inbox.

Hospitality
Hospitality

Same five questions, every single day.

Hours, bookings, dietary options, parking — the Support Agent demo further down handles the repeat questions so staff handle the room.

Page 1 / 6 — click or swipe to flip

How I approach this

I don’t start with “what AI tool should we use.”

I start with: what happens when this breaks, slows down, or gets missed?

Built systems

Proof, not a pitch.

Three demo builds you can actually run. Each one is a prototype built to show how the underlying system works — not a claim about a real client project.

Interactive PrototypeSimulated data — no live backend
Waiting

AI Lead Recovery System

See what happens when a new lead arrives — even after hours. Instead of sitting in an inbox until tomorrow, the system qualifies the lead, updates the CRM, drafts a response, alerts the sales team, and schedules follow-up.

Demo enquiry

Sarah Mitchell — Mitchell Renovations

“Hi, we're looking for a quote for renovating our kitchen. We'd like to get someone out to look at the property next week.”

Without automation

  1. New enquiry
  2. Sits in inbox
  3. Someone notices it
  4. Someone qualifies it
  5. Someone enters it in the CRM
  6. Someone responds
  7. Someone remembers to follow up

With automation

  1. New enquiry
  2. AI qualification
  3. CRM updated
  4. Personalized response
  5. Sales alert
  6. Follow-up scheduled

Also built

Demo Build 02Simulated data — no live backend
Waiting

AI customer support agent

Support inboxes fill up with the same handful of questions. This agent reads the request, pulls the real order/account context, and drafts a reply — instead of a human retyping the same answer for the tenth time today.

Input
AI / Logic
Automation
Output

Try a question

Demo Build 03Simulated data — no live backend
Waiting

AI business admin automation

Every incoming email that needs sorting, logging, and a reply eats time that should go to actual work — not retyping the same three responses.

Input
AI / Logic
Automation
Output
Incoming email received
AI reads email
Extracts key information
Categorizes request
Updates spreadsheet/database
Creates task
Sends response

What I build

Three ways automation actually earns its keep.

A

AI Agents

Conversational and task agents wired to your actual tools — not a chatbot that stops at answering questions.

W

Workflow Automation

n8n pipelines that move data and trigger actions across the tools you already run, with errors that surface instead of failing silently.

S

Systems Integration

APIs, webhooks, and data plumbing connected so your existing software actually talks to each other.

Who this is for

The pattern repeats everywhere leads or admin work pile up.

The three demos above map to real categories — an inbound-lead business, a support inbox, a repetitive admin process. That pattern shows up across more industries than just renovation.

Trades & Home ServicesReal EstateE-commerce & RetailProfessional ServicesHospitalityHealth & Wellness

Stack

Real tools, not buzzwords.

AI & Agents

Claude

Automation

n8n

Product

Next.jsTypeScriptTailwind

Infrastructure

VercelGitHub

Testing

Playwright

Build proof

This website is proof #1.

Jayson AI Holland — this website

Built with Claude Code, in the open — the Build Log documents every real iteration.

More builds, coming

Additional repositories, live demos, and architecture writeups get linked here as each one ships — nothing invented ahead of time.

Have a process worth automating?

Tell me what’s happening today — I’ll tell you straight whether automation actually helps.

Start a build