Brandon Chew · Singapore
What I actually do
for a living
Pre-sales Solutions Architect in AI. Nobody grows up wanting to be one. Explained via 60kg bars, koyok, and demo monkeys.
play as slides instead →The one-liner
Two skills, married:
technical aptitude × talk and sell.
Half engineer, half translator, full-time matchmaker between what a product does and what a customer actually needs.
Identity crisis
One job, nine names.
Keyword: pre-sales. Trap: one famous AI company flips it: "solution engineer" = pre-sales, "solution architect" = post-sales. Read the JD.
The job, in six acts
What a day looks like
Discovery
Find the customer's real pain; price it in hours and dollars, not vibes.
Demos
Story built on their pain: solution, not feature tour. Years to learn.
Proof of concept
Demo says "it can work." PoC proves it: their data, their security, their environment.
Solution design
Requirements become architecture post-sales can actually deliver. Nobody inherits fiction.
RFPs & tenders
One of the most irritating things ever invented. Good news: AI helps fill them. Division of labour, settled.
Voice of the customer
The customer's advocate in-house, lobbying product teams for the roadmap the market actually wants.
And through all six: trust. Win it across the table, win the sale. Trust is the product before the product.
Expectation management 101
"They asked to deadlift 200kg on day one.
I handed them the 60kg bar."
First you find out where the ask is coming from. Sometimes the real job is moving a piano, and you don't need a deadlift at all. Then: what's solid today, what's honestly on the roadmap.
The golden rule
Don't sell koyok.
Koyok = snake oil. Over-promise and the blame lands on post-sales, not me. And post-sales will come find me.
I hand over like a surgeon to the ward: clean, honest, zero surprises.
Plot twist
You already sell. Every. Single. Day.
Wedding plans, to your parents? Selling.
New drink, for the same-order friend? Selling.
One more episode of the favourite show, at 1am? Selling.
Not an alien skill. I just get paid to stay laser-focused on it.
Myth-busting
Two legends about my job
❌ "You're just a demo monkey"
Mostly legend now. Happens when a junior AE treats the SA as a button to press. Good AE+SA = partnership, not summons.
❌ "You must know everything"
False: nobody knows everything. Resourcefulness beats omniscience. Curveball comes, stay calm, cook an answer path on the spot.
Almost forgot
Confession: the worst computer science student.
My programming sucked. I could always see the bigger picture. Apparently that's the whole job. Career path: integrations & automation → search & observability → AI.
What I tell every mentee: loving the product you sell beats the payslip logo. Wrong cup of tea? Your body tells you first.
In cold, hard numbers
How the job pays (and why it's safe-ish)
(harder product = bigger carrot)
than pure eng roles.
my way into AI pre-sales. Every day.
The AI question
"Isn't AI coming for your job?"
Literally debated this on stage at Pre-Sales Collective: "AI will replace pre-sales in 5 years." My side won. The keyword is replace vs change. AI automates about a third of the work (McKinsey's own math). Two-thirds stays human.
75%
of B2B buyers will prefer human interaction over AI by 2030. AI caused the backlash itself. (Gartner, Aug 2025)
1.65×
more buyer's remorse on digital-only purchases. Solo buyers crash; reps + tools are 1.8× more likely to complete a high-quality deal. (Gartner, B2B Buying Report)
The ATM lesson
ATMs didn't end bank tellers. They ended ledger-typing and made relationships the job. AI ends my typing, not my trust. (James Bessen, Learning by Doing)
The AI question, continued
What AI actually did to my job
✅ AI eats the boring third
Demos: two a day, now four. AI writes RFP first-drafts, researches customer KPIs and news, assembles my narrative.
🚫 What it can't read
Body language. Offline politics. The WhatsApp call that never makes the transcript. When AI answers wrong, someone still owns it publicly. Air Canada's chatbot found that out the hard way.
"AI won't replace humans — but humans with AI will replace humans without AI."
— Karim Lakhani, Harvard Business Review
The honest part
Yes, I've burnt out before.
Three strikes. Three customers pissed me off; young me took it personally and lost weeks of sleep.
Now? "You want to be nasty? Okay lor." Skeptical customers keep me sharp; boundary-less rudeness gets no rent-free headspace. Emotional resilience is a learned skill, like demos.
So, why do it?
"When I'm doing the work,
it doesn't feel like work."
Long hours. Evil RFPs. But every week: new industry, new problem, new room to win over, plus eight years of lessons for the next mentee.
That's the job