Deciding to use AI is the easy part. Deciding who to trust with it is not.
If you run a shop and you are looking for an AI agency in Indonesia right now, you have probably noticed the same thing we have. Almost everyone suddenly does AI: web shops, marketing agencies, freelancers, even brand-new startups. The label is everywhere, but the quality underneath it varies a lot.
This guide is here to help you tell the difference. It is written to be useful even if you never talk to us at all. We just want to give you the questions to ask, so you can hire an AI agency in Indonesia with your eyes open.
What a good partner should actually do for a shop
Set the buzzwords aside. A good partner does three ordinary things well:
- Finds the right problem to solve first: not the flashiest one, the one that pays. (We wrote a whole piece on where AI actually pays off first.)
- Builds the smallest thing that proves it works, before asking you for a big budget.
- Tells you the truth, including when the honest answer is "you do not need us for this."
If an agency cannot clearly explain how it does those three things, the rest of the conversation does not matter much.
The questions to ask before you hire
Bring these to any agency you are considering. The answers tell you more than any portfolio.
1. "Do you start with my shop, or your tool?"
A good partner spends the first conversation asking how your business runs: where orders come from, how stock gets counted, what happens on a busy weekend. If they start recommending tools before they understand how you make money, that is a warning sign. This kind of listening should always come first. It is exactly what our health check is built around.
2. "What number will this move, and how will we measure it?"
Real work is measurable: hours back each week, fewer stockouts, faster replies on WhatsApp, fewer orders typed in by hand. If you cannot agree on a number before the project starts, you will never agree on whether it worked. Vague promises about "efficiency" are a polite way of avoiding accountability.
3. "Can we start with something small?"
Be careful with anyone who wants a large, long, upfront commitment for a first project. A confident partner will happily prove value in a short, contained piece of work first. Small is not a lack of ambition. It is how you lower your risk.
4. "Who owns the code, the data, and the accounts?"
This one gets skipped, and it matters a lot. You should own your code, your sales data, and the accounts everything runs on, including the marketplace and payment connections. Ask directly. If the answer is unclear, or everything lives inside the agency's accounts, you are renting your own business back from them.
5. "Will you tell me when AI is the wrong choice?"
This is the sharpest question you can ask. A good agency has turned down work before. Sometimes a tidy stock count, a feature you already pay for, or an extra pair of hands works better than a model, and a partner worth having will tell you so. If everything you describe gets a yes, they are selling, not advising. (Here is when we think AI is not the answer.)
6. "Have you worked with a shop like mine?"
Retail has its own plumbing, and it breaks in its own ways. Ask whether they have actually connected Shopee, Tokopedia or TikTok Shop to a real stock count, whether they have worked with a POS like Moka or Majoo, whether they have set up QRIS or a payment gateway for someone. Ask what they do when a marketplace changes its API and orders stop arriving on a Saturday.
An agency that has never done this will learn it on your money, and you will find out during Harbolnas.
7. "Can you support us here, in our language and timezone?"
This is where a local agency has a real edge over a distant vendor. You get the same timezone, both Bahasa Indonesia and English, and people who understand how business actually runs here. That is not a small thing when checkout breaks at 4pm on a Friday.
Red flags to walk away from
- Guaranteed results before they understand your shop.
- A chatbot pitched before anyone has looked at whether your stock numbers are right.
- Prices with no scope: a big number and a vague promise.
- No interest in measuring anything.
- Pressure to sign fast.
- They cannot explain, in plain words, what they would actually build. If they are proposing AI agents, they should be able to tell you exactly what those agents will and will not do.
Green flags worth paying for
- They ask more questions than they answer in the first call.
- They propose a small first step.
- They are comfortable telling you what they are not good at.
- They can point to real work they have shipped.
- They talk about your numbers, not their technology.
None of this is unique to AI. It is just good advice, plainly given. If you want a wider view on strategy, Harvard Business Review's writing on AI is a level-headed place to read more.
How we try to answer these
We built FocusBinary to be the kind of partner we would want to hire for a shop. That means starting with your problem, proving it small, measuring one number, and being honest about the limits.
That is also why our work runs in steps instead of all at once: get online, connect your apps and stock, automate the busywork, and add AI once the rest holds. Most shops are not on the last step yet, and we will tell you so.
If you want to put these questions to us directly, get in touch. Bring the hard ones. That is the whole point.
