AI · Systems · Automation

Complex problems.Smarter shortcuts.

i’m Skye, an AI engineer in Singapore. i build speed-to-lead systems: agents that answer every enquiry in seconds, qualify it, and book it before the lead goes cold.

skye@thesmartshortcut.com  ·  +65 9363 1141

The leak

A lead doesn’t go coldin a week. It goes coldin five minutes.

the person who messages at 10pm is the one who is ready to book. by morning they’ve found someone else.

most businesses don’t have a lead problem. they have a response time problem. it stays invisible because the lead never complains, they just quietly go elsewhere.

What I build

Three systems. One job:stop the leak betweeninterest and booking.

  • 01

    Speed-to-lead systems

    Every enquiry answered in seconds, on the channel your customers already message you on. Qualified, routed and logged before anyone on your team opens a laptop.

    Instant replyQualificationRoutingHandoff
  • 02

    AI customer agents

    An agent that reads your documentation and holds a real conversation. Questions, availability, scheduling, bookings. It stays inside what your business actually knows, and hands to a human the moment it should.

    Natural languageBookingsKnowledge groundedEscalation
  • 03

    GTM engines & AI second brains

    Outbound that sources, researches, writes and sends on its own. Internal knowledge your team can query instead of digging through folders. Built to run without anyone watching it.

    OutboundEnrichmentKnowledge baseDashboards

Proof

One client. One month.Numbers I can defend.

7

Booked calls

from those enquiries

33

Enquiries handled

in one month

21.2%

Booking rate

enquiry to call

$125K

SGD pipeline

at her list price

Case 01

A creative agency serving real estate clients

Singapore  ·  Built on Crumbs

She was getting more enquiries than she could answer, and the slow ones were dying in her inbox. We stood up an agent on the channel her clients already used. It replied in seconds, asked the qualifying questions she would have asked, and handed over a warm, sorted lead.

33 enquiries in a month. Every one answered and qualified. 7 turned into booked calls.

Said plainly: the $125K is pipeline at her standard package rate, not closed revenue, and it’s her revenue, not mine. One client, one month. I’d rather show you a number I can stand behind than a number that sounds better.

Case 02

A day spa in Bali

Design & build  ·  Not deployed

Designed and built a booking agent for the owner: answers the questions guests actually ask, checks what’s available, and walks a conversation to a booking in plain language rather than a form.

It was built and specced but never went live, so there are no results to quote. It’s here because it shows the work, not because it proves an outcome.

How it goes

Proof before platform.I learned that onethe expensive way.

  1. 01

    Find the leak

    One call. I look at where enquiries arrive, how long they wait, and where they stop. Usually the number is worse than anyone expects, and that's the useful part.

  2. 02

    Build the smallest thing that fixes it

    Not a platform. The one agent or workflow that closes the specific gap, grounded in what your business actually knows, running on the channel you already use.

  3. 03

    Watch it in the wild, then tighten

    Real enquiries, real conversations, measured. I tune it against what people actually ask rather than what we guessed they would.

  4. 04

    Hand it over

    You own it. Documented, and simple enough that you don't need me on retainer to change a line.

Who you’d be working with

Saffuan Skye.

30, based in Singapore. Before AI I was an assistant product owner at Manulife Singapore, working on ManulifeMOVE. CSPO and CSM certified.

I came into this from the product side, which mostly means I care more about whether a thing gets used than whether the architecture is elegant.

I also built a platform before I had proof anyone wanted it. That was an expensive lesson and it’s the reason I now start every engagement by finding the leak rather than scoping a build.

Trust is the hard part, not the replies

Getting a model to answer is easy. Getting it to stay inside what your business actually knows, never invent, and know the moment to hand to a human is the whole job.

Not everything should be automated

Some things are meant for humans. The skill is telling which is which, and being willing to say so when the answer costs me scope.

People don't pay because they can't

They pay because they don't have the time, energy or headspace to do it consistently. Build for the consistency, not the capability.

Start here

Tell me what’sleaking.

One call, no deck. If I can fix it I’ll tell you how, and roughly what it takes. If I can’t, I’ll say that too and point you somewhere better.