We work out what you should do first

A 30-minute free diagnostic and a one-page report: where you are, three viable entry points, the recommended next step. The audit, PoC and rollout are each priced separately, and you can stop after any of them.

perhapxin@perhapxin.com See price ranges

Three things we hear a lot

The place people get stuck is almost always the same, and it has nothing to do with technology.

No idea where to start
- You can list what you want to do; you can't rank it.
- Every item looks like it should happen.
- Nobody has worked out which one pays off first.
Not sure it's worth it
- How long until it shows a return — you don't have that estimate.
- How many person-hours this costs you a year — never counted.
- What getting it wrong would cost — counted even less.
You can't get a judgement
- You ask a vendor and get a quote back.
- The quote doesn't tell you whether to do it at all.
- What you want is someone to think the problem through with you.

This is where most Taiwanese SMEs are

92%

Know AI only “roughly”

Over nine in ten SMEs have heard of it and broadly understand it — no further.
46%

Planning, not started

Far more companies intend to act than have actually begun.

Four questions that set the order

We ask the same four on every engagement. The answers decide what goes first, how much validation it needs, and whether training belongs in scope.

What does this cost you in people today?
- Start with the denominator: how often, how many people, how long each time.
- Daily, repetitive work eating half a person goes first.
- A process that runs twice a year goes last.
What does getting it wrong cost?
- Anything touching customers, money or compliance gets validated with a small PoC first.
- Internal-only impact can be built directly and adjusted as you go.
- Naming the cost is what tells you how much validation is enough.
Is the data there, and is it clean?
- Which systems hold it, and whether they have APIs.
- Formats, gaps, and how much cleanup comes first.
- Where the data isn't ready, we'll say don't start yet.
Can your team carry it?
- Whether anyone internal can take over operations.
- If not, training and written procedures go into the scope.
- This one decides whether you'll keep having to come back to us.

After the four questions, cases fall into three groups

A diagnostic usually surfaces three to five candidate scenarios. Once the four questions are answered, each one lands in a group below — and that decides whether it goes on to a quote.

Ready to build now
- The data is already in a system and reachable.
- The process hasn't changed much this year.
- The time saved converts into a number.
How we handle itStraight into a PoC: 4–6 weeks to turn that one scenario into a testable prototype.
Fix the data first
- The scenario is right, but the data lives on paper, in Excel, or in one person's head.
- The same record doesn't reconcile between two systems.
- The fields exist; half of them are empty.
How we handle itSpend 2–4 weeks consolidating sources and formats first. Skip this and whatever gets built afterwards won't be accurate either.
Don't start yet
- The process itself is still changing month to month.
- The manual volume was never large to begin with.
- What's actually stuck is unclear ownership.
How we handle itThe report will say not yet, and what we suggest fixing first. We won't take this one.

Pricing

Each stage is priced on its own. The actual figure is confirmed in a quote once the scope is clear.