Book a call
A founder's manifesto · Singapore

Most AI projects fail before the first line of code.

I started RightClick:AI because most automation fails for the same reason — and it is not the technology. The work that decides whether a system survives contact with a real business is process design, data mapping, and clarity before tools. Most agencies skip it to ship faster.

This page is what I do instead, and what I refuse to compromise on.

Five things I refuse to do.

These are not preferences. They are non-negotiables. I have said no to projects worth more than the cost of this entire website because the engagement would have violated one of them. The list is short, and it is the reason the work stays good.

01

I will not pick a tool before I understand the workflow.

The technology decision is the smallest decision in the project. If the workflow is mapped poorly, no tool — Zapier, Make, n8n, custom build — will save it. Most automation projects skip this and pay for it later, usually at the worst possible moment.

02

I will not take a project where the data cannot support the decision.

You cannot automate what you cannot measure. If the data does not exist, is unreliable, or cannot be cleanly extracted, I say so before the contract is signed. The honest answer is sometimes "fix the data first, then automate."

03

I will not bill a client for a system I would not run my own business on.

Every system I ship, I evaluate against the same standard I evaluate my own internal tools by. If I would not use it, I do not deliver it. The work is too expensive on both sides to be embarrassed by the result.

04

I will not promise a timeline I cannot back with a dated refund clause.

Six weeks means six weeks. Every engagement comes with a written timeline and a SGD 200/day credit for missed deadlines. If I cannot meet the date, you do not pay full price. The clause exists because soft promises are how trust dies.

05

I will not hide behind an account manager.

The person who scopes the work is the person who builds it. No briefing chains. No junior developers interpreting senior promises. You work directly with me at every stage of the engagement — the mapping session, the prototype, the build, the handover.

Where growing service businesses get stuck.

Service businesses between fifteen and thirty-five people tend to hit the same three structural bottlenecks. They are not strategy problems and they are rarely about the team's capability. They are operational frictions that compound quietly until the founder is the slowest part of the system.

Your operation · This month
Client report assembly
14 hours this week. Same as last week.
Manual
"Where are we with Sarah?"
Asked 3 times. 4 systems opened.
Scattered
Approval queue
6 items waiting. Average idle: 3 days.
Stalled
Internal status update
Slack thread. Tuesday. Still unread.
Lost
Cross-team handoff
Designer → PM → Client. Two days each gap.
Idle
  • 01

    Reporting overhead.

    Producing client reports, internal status reports, and management updates absorbs an enormous amount of skilled time. Most of it is manual data assembly the team would never choose to do if the system did it for them.

  • 02

    Fragmented client visibility.

    Information about a client lives across email threads, project tools, finance software, and individual heads. Answering a basic question — what stage is this client at, what is outstanding, what is at risk — requires asking three people and opening four systems.

  • 03

    Approval and handoff friction.

    Work waits on someone to be notified, to remember, to approve, or to pass it along. Each handoff loses time and context. The slowest part of delivery is usually not the work itself — it is the gap between the work.

There is a quieter problem underneath all three: the parts of the operation that fall between roles. The reports nobody owns. The approvals nobody chases. The data nobody is responsible for keeping current. Those gaps grow as the business grows, and they are where founders end up spending their attention by default.

The conventional answer is to hire. Hire an operations manager, hire an analyst, hire a chief of staff. That works, but it is expensive, slow to onboard, and adds another node that needs to be managed. I am proposing a different answer.

Data Process Thinking.

Most automation projects fail not because the technology was insufficient — but because the process was never clearly understood before the build began. The bottleneck is not technology. It is clarity of process.

Design before you build. Build before you scale. The four steps below are non-skippable. The first three are where I do most of the work.

01
Map input → output at every stage

Every workflow has explicit inputs and outputs. We map them granularly — who, when, with what data, in what format. The map becomes the specification.

02
Surface the invisible micro-processes

The small decisions people make without thinking are the ones that break automated systems. We name them all before the build, not after.

03
Verify the data behind the decision

You cannot automate a decision the data cannot support. We check what exists, how reliable it is, and whether it can carry the system. If it cannot, you hear it before the build starts.

04
Only then: design and build

With the map, the data, and the edge cases clear, the build is straightforward. Six weeks, working prototype in three. No surprises, because we eliminated them up front.

"

Clarity before tools.

Bolun (Ben) Liu, Founder

Bolun (Ben) Liu

Founder, RightClick:AI · Singapore
Published 2026-05-09

I am a marketer and an AI operator. I run AI workflows inside a service-based SME by day, and I started RightClick:AI because I watched too many service businesses pay for automation that never produced what was promised — not because the technology failed, but because nobody mapped the process before anyone wrote any code.

My edge is not code. It is process-first thinking with data discipline. Before I design any system, I map every input and output. I identify what data exists and whether it can support the decision the system is supposed to make. I surface the small invisible micro-processes that nobody consciously tracks. Only then does the build begin.

Everything else on this page — the manifesto, the method, the engagement terms — is the operating system I work by. I have not changed my mind about any of it.

Sectors delivered: healthcare consulting · financial services · compliance workflows · family office operations · 3PL logistics · education consulting · marketing agencies · law firms.
Contact: ben@3nm.io · Singapore (SGT, UTC+8) · 24-hour response.
For partners

I work with partners who own the client relationship — agencies, consultancies, advisory firms. The engagement is signed in your name. Your client never sees a "RightClick:AI" invoice unless you want them to. I am the backend; you stay the brand.

If you've read this far

If this resonates,

write to me directly: ben@3nm.io.

I read every email. I reply within 24 hours.

If it's a fit, we'll talk. If it isn't, I'll tell you why.

— Ben
Bolun (Ben) Liu · Singapore (SGT, UTC+8)