● AI Automation Engineer · AI Systems Builder for business operations

I design and build AI-powered automation systems that remove repetitive operational work.

I build around the way your business already runs. I connect the tools you’re already using, plan for the cases where things go wrong, and make sure the team that depends on it can actually understand it.

fig. 01: the main path, plus where it hands over
11 systems documented3 in real useexceptions + handoffs shown for each one

01 Sound familiar?

The work that quietly eats your week.

None of these is a big dramatic problem. They just keep happening, every day, and somebody on your team is always the one covering for them.

  • ⚠ 01

    Somebody is copy-pasting all day

    Data moves from email to sheet to CRM to chat by hand, and one wrong paste breaks everything downstream.

  • ⚠ 02

    Replies depend on who is free

    Customers wait because the person who handles bookings or enquiries is busy with something else.

  • ⚠ 03

    Follow-ups that never happen

    Everyone agrees the follow-up matters. It still gets skipped every busy week.

  • ⚠ 04

    Information in five places

    The order is in one place, the status in a chat, the customer details in a sheet, and nobody has the full picture.

  • ⚠ 05

    Nobody knows when it breaks

    Some automation already exists, but when it fails, it fails quietly, and you find out from an angry customer.

  • ⚠ 06

    The process lives in one head

    One person knows how it all works. When they’re away, everything slows down.

02 What I build

Systems for specific parts of the business.

I don’t sell tools. I look at one workflow, work out where it breaks, and build the system that runs it properly.

  • Lead Qualification & Appointment Booking

    People message you to book and somebody has to reply, check the calendar, go back and forth, then remember to confirm. Some of them just stop replying because nobody answered fast enough.

    intent detectionlive availabilityreschedule / cancelhandoff to staff
  • Sales & Outreach Automation

    Prospect lists live in a sheet, research is manual, and every email is either copy-pasted generic or takes ten minutes to write.

    lead sourcingvalidationqualificationdraft + review
  • Customer Support Automation

    The same questions keep coming in, the answers live in somebody’s head or a scattered doc, and urgent issues get buried under the easy ones.

    grounded answersconfidence fallbackescalationticket updates
  • E-commerce Operations Automation

    Orders, riders, delivery updates and returns are tracked across chats and spreadsheets, so customers ask “where is my order?” and nobody is sure.

    order recordsrider assignmentcustomer updatesreturns & refunds
  • Recruitment & HR Automation

    Applications come from everywhere, candidate status lives in someone’s inbox, and recruiters spend more time updating trackers than talking to people.

    candidate recordsstagesrecruiter actionspipeline view
  • Document & Back-office Automation

    Invoices and documents are read and re-typed by hand, matched against POs by eye, and approvals get stuck in someone’s inbox.

    extractionmatchingapproval routingmismatch exceptions
  • Custom AI Internal Tools

    Your team needs a small tool that fits exactly how you work, and off-the-shelf software is either too much or not quite it.

    internal appsknowledge assistantsrole-based actions

What each system actually does

03 Selected work

Things I’ve built, and what state they’re in.

Every project has an honest status label. Some are running for clients, some are completed builds, some are prototypes. I’d rather you know which is which.

  • 01Production client system

    WhatsApp Appointment Booking Automation

    Customers book, reschedule and cancel appointments by chatting on WhatsApp, and the calendar stays correct without anyone babysitting it.

    1. WhatsApp message
    2. Understand intent
    3. Check availability
    4. Book the slot
    5. Send confirmation
    6. +2 more
    4 failure paths · 3 human handoffs
  • 02Workflow build / implementation

    E-commerce Order-to-Delivery Automation

    Every order gets one record that follows it from fulfilment to rider to doorstep, and the customer hears what is happening without having to ask.

    1. New order
    2. Order record created
    3. Fulfilment
    4. Rider assigned
    5. Delivery update
    6. +1 more
    3 failure paths · 3 human handoffs
  • 03Workflow build / tested scenarios

    E-commerce Return Request Automation

    Return requests are checked against the policy, recorded, and walked through to refund or replacement, with invalid ones going to manual review.

    1. Return request
    2. Eligibility check
    3. Return recorded
    4. Instructions sent
    5. Return tracked
    6. +1 more
    3 failure paths · 3 human handoffs
  • 04Client workflow / pilot

    Dental Patient Reactivation System

    Patients who are overdue for a visit get a personal follow-up with a booking link, and the clinic can see who booked from one Airtable view.

    1. Identify overdue patients
    2. Personalised follow-up
    3. Booking link
    4. Booking tracked
    5. Clinic sees results
    3 failure paths · 2 human handoffs
  • 05Completed workflow

    AI Cold-Outreach Draft Helper

    Each prospect row in a sheet gets qualified and a personalised draft written back next to it, ready for a human to review and send.

    1. Prospect row
    2. Qualification
    3. Personalised draft
    4. Written back to sheet
    5. Human reviews & sends
    3 failure paths · 3 human handoffs
  • 06Deployed application

    RecruitTech ATS

    A deployed applicant tracking app where every application becomes a candidate record moving through screening and interview stages.

    1. Application
    2. Candidate record
    3. Screening / status
    4. Interview stage
    5. Recruiter action
    6. +1 more
    3 failure paths · 2 human handoffs

See all 11 systems

04 From messy to reliable

Same work. Different system.

This is the WhatsApp booking system I built for a client. Switch between how it ran by hand and how it runs now. Click any step, or walk through it, and look at what happens when something goes wrong.

How it ran manually · depends on whoever is free

  1. 01Customer sends a message on WhatsApp asking to bookby hand
  2. 02Someone reads it whenever they get a chanceby hand
  3. 03They open the calendar and check what is freeby hand
  4. 04Back and forth with the customer until a time worksby hand
  5. 05Booking is entered by hand, confirmation typed by handby hand
  6. 06Reschedules and cancellations start the whole thing againby hand

Read the full case study

05 How I work

Understand it first. Then build it.

Most automation goes wrong because somebody started building before they understood the process. I do it the other way round.

  1. 01Discovery
  2. 02Workflow mapping
  3. 03System design
  4. 04Build / integration
  5. 05Testing
  6. 06Go live
  7. 07Handover / monitoring

What happens at each step

06 Reliability

The happy path is the easy part.

Any demo works when the data is clean. What matters is what happens on a bad day, so the failure paths and handoffs are part of the build from the start.

  • Exceptions are designed, not discovered

    Missing data, unclear messages, API failures, no availability. I map these before building, and each one gets a defined path.

  • People step in at the right moments

    Approvals, judgement calls and anything sensitive go to a person, with the context already attached so they don’t start from zero.

  • Failures are loud

    If something breaks, someone gets told. A workflow that fails quietly is worse than no workflow, because you think it ran.

  • One source of truth

    Every system writes back to a record the team can actually see, so the automation is never a black box.

Try it · failure scenarios from the booking system

Where it happens
Step 02 Understand intent
What goes wrong
Message is unclear or not about booking
What the system does
The bot asks a clarifying question instead of guessing, and if it still can’t tell, the chat is flagged for a staff member.
Who handles it
A person: staff
Portrait of Hubroot

07 About

Hi, I’m Hubroot.

I build automation systems for business operations. Booking flows, outreach pipelines, order tracking, returns, support assistants, small internal tools. Most of them start with somebody on a team doing the same thing by hand every day.

I care more about the system working on a bad day than looking good in a demo. So I spend a lot of time on the boring parts: what happens when data is missing, when an API fails, when a customer says something the bot doesn’t understand. Those are the bits that decide if people trust it.

More about me

08 Questions

Before you ask.

01Do I have to change the tools I already use?

Usually no. I build around what you already have where possible: your sheets, CRM, WhatsApp, email, calendar. If one of your tools genuinely can’t be connected, I’ll tell you that early instead of finding out halfway.

02What if my tool doesn’t have an integration?

Most tools have an API or webhook of some kind, even if there’s no ready-made connector. If there really is no way in, we look at alternatives together, but I won’t pretend something connects when it doesn’t.

03What happens when the automation breaks?

Things will fail sometimes. APIs go down, people send weird data. The difference is whether the system was designed for it. I build failure paths, alerts and human handoffs into the system so problems are visible and handled, not silent.

04Will the AI make things up or say something wrong to my customers?

I only use AI where it helps, and I keep it on a short leash: grounded in your information, with clear fallbacks and handoff to a person when it isn’t sure. For anything sensitive, a human approves before it goes out.

05Do you offer support after the build?

Yes. I hand over with documentation and stay around to monitor and fix things as the system runs in real conditions. We can agree on ongoing support depending on what your setup needs.

06Is my process a good fit for automation?

If it’s repetitive, rule-based (mostly), and happens often enough to be annoying, probably yes. If it’s a one-off, or every case needs deep judgement, maybe not, and I’ll say so. Tell me about it and we’ll figure it out.

07How long does a build take, and what does it cost?

It depends on the workflow, the number of tools involved and how many exception paths there are. After I understand the process, I’ll give you a clear scope before anything starts.

Next step

Tell me what you’re trying to automate.

Describe the workflow in your own words. What happens now, where it hurts, which tools are involved. I’ll read it and get back to you.

Tell me what you’re trying to automate