Choosing an AI Partner: Seven Questions to Ask
Everyone offers AI services now. 🎭 Firms that were building websites last year and running social media the year before are selling “AI solutions” today. Some genuinely deliver; others changed the sign above the door.
Telling them apart looks hard but isn’t. Ask the right questions and the difference surfaces in the first meeting — because people who know the work talk differently. 🔍
This guide gives you those questions: what to ask, which answers are good, which signals are bad and how to compare proposals. 📋
We provide these services too, so let’s say it plainly: this list applies to us as well. Ask us the same questions; if there’s somewhere we can’t answer, you have a right to know. ⚖️
Seven Questions to Ask ❓
BU BÖLÜMÜN ÖZETİ
- “Which of my tasks will you automate?”
- “Why should this task be automated?”
- “How long until I see results?”
- “How will you teach my team?”
- “What happens if it produces a wrong output?”
- “Where will my data live?”
These seven reveal what the other party actually does within an hour. None require technical knowledge.
We’ve written what a good answer looks like too. ✅
“Which of my tasks will you automate?”
A good answer begins with a question: “I’d need to look at your operation first.” 🎯 A bad one immediately proposes a tool — anyone proposing a solution before understanding the work is selling the solution. The correct order is settled: work first, tool second.
“Why should this task be automated?”
A good answer discusses the time and cost to be saved. 💰 A bad one says “your competitors are doing it” — that’s pressure, not a rationale. Investments made because a competitor did tend to start unmeasured and end unmeasured.
“How long until I see results?”
A good answer is staged: setup takes weeks, adoption takes months. ⏳ Anyone saying “ready in a week” is either describing something very simple or overselling. The distinction matters: a week may be accurate for simple work, but the price should reflect that.
“How will you teach my team?”
This question is the most revealing of the seven. 🎓 A good answer contains a concrete plan; a blank look means they intend to hand over the build and leave. An untaught system goes unused, as covered in our cost guide: this item puts the entire budget at risk.
“What happens if it produces a wrong output?”
A good answer describes the review mechanism. ⚠️ Anyone saying “it won’t” is either inexperienced or not being straight — every system can be wrong. The real question: who notices, and how does it get corrected?
“Where will my data live?”
A good answer is specific and technical: which service, which setting, which region. 🔐 Anyone brushing the question aside hasn’t considered it — and the responsibility ends up with you, not them.
“What happens if I stop working with you?”
The critical question. 🚪 Does the system stay with you, whose name are the accounts in, will the process be documented? A vague answer means dependency is being built. The best time to ask is the first meeting, not the day you want to leave.
Warning Signs 🚩
BU BÖLÜMÜN ÖZETİ
- Proposing before understanding
- No concrete numbers
- No sense of limits
- Creating urgency
Five signals indicate the other party is sales-driven. All are visible in the first conversation.
Even one warrants caution. ⚠️
| Signal | What it means |
|---|---|
| Proposes without asking about your work | Selling a product, not solving a problem |
| Gives no concrete numbers | Hasn’t thought about measurement |
| Claims anything can be automated | Doesn’t know the limits |
| No references, or can’t show them | You may be the first attempt |
| Creates urgency | Doesn’t want you comparing |
Proposing before understanding
If a tool name comes up in the first meeting, be careful. 🔨 The order should be: understand the work, then choose the tool — the reverse is trying to sell what they already have. A firm tied to one tool ends up proposing the same solution to every problem.
No concrete numbers
“We’ll improve your efficiency” is a sentence, not a commitment. 📊 A good proposal estimates how many hours will be saved on which task and offers to measure that estimate. The estimate may miss; what matters is that it was made.
No sense of limits
“We can automate anything” means they don’t know the boundaries. 🚫 An experienced partner also explains what can’t be automated.
Creating urgency
“This price is valid this week only”. ⏰ A good partner wants you to compare, because they aren’t afraid of being chosen after comparison. A proposal that rushes you is one that knows it looks weak under scrutiny.
Good Signs ✅
BU BÖLÜMÜN ÖZETİ
- They ask first
- They set limits
- They propose starting small
- They raise the exit themselves
Four signals indicate the other party genuinely knows the work. You rarely hear these in a sales conversation — which is exactly why they matter.
Even two makes for a good candidate. 🎯
They ask first
In the first meeting they mostly listen and you mostly talk. 👂 Someone who knows a solution can’t be built without understanding the work spends time asking. A rough test: if you spoke for more than half the meeting, you’re on the right track.
They set limits
Someone who says “we can do this, we can’t do that” is reliable. ⚖️ Knowing your limits means keeping your promises; someone with no limits goes quiet at the first difficulty.
They propose starting small
Those who know the work suggest starting with one task. 🎯 Anyone proposing a transformation package is either inexperienced or aiming at your budget.
They raise the exit themselves
The best signal of all. 🚪 The system stays with you, accounts in your name, process documented — a partner who says this before you ask isn’t planning to build dependency. They’ve chosen to secure continuity through quality rather than lock-in.
Comparing Proposals 📊
BU BÖLÜMÜN ÖZETİ
- Scope: what’s in, what isn’t
- Timeline: when it ends
- Measurement: how success is defined
- Handover: what happens afterwards
The difference between two proposals is scope, not price. Two proposals that look alike can describe very different work.
Compare on four headings. 🧮
Scope: what’s in, what isn’t
Is setup included, is training, is maintenance? 📋 The cheaper-looking proposal has usually left one of them out — and that item arrives later. When comparing, add the missing items yourself and recalculate the total.
Timeline: when it ends
How many weeks until setup completes? ⏱️ A proposal without dates is a project without an end; work with no completion date tends to stretch.
Measurement: how success is defined
Does the proposal state which number will improve? 📈 If not, whether the project succeeded stays a matter of opinion at the end.
Handover: what happens afterwards
Whose name are the accounts in, will the process be documented, what’s delivered if you leave? 🔑 These belong in the contract — discussed later, they become a negotiation.
Where We Stand 🤝
BU BÖLÜMÜN ÖZETİ
- We say when it isn’t needed
- We start small
- The system stays yours
Let’s state our own position so you know what to expect. We commit to three things.
They get repeated in the discovery call. 📋
We say when it isn’t needed
If your processes aren’t documented, your scale is small or your real problem sits elsewhere, we’ll tell you. 🎯 A project started at the wrong time damages both sides.
We start small
The first engagement is a single task. 🔧 If four weeks produce no measurable result, widening the scope makes no sense — widening magnifies the mistake too. If nothing comes out, we look for the cause together; that’s an output as well.
The system stays yours
Accounts are opened in your name, the process is documented, and it keeps running when we leave. 🔑 Scope and method: AI Consultancy. To see where you stand today: Digital Audit. 🚀
Frequently Asked Questions 💬
Sık Sorulan Sorular
Seven questions: which task, why, how long until results, how the team will be taught, what happens on wrong output, where data lives and what happens if I leave.
“What happens if I stop working with you?” Does the system stay with you, whose name are the accounts in? A vague answer means dependency is being built.
Five: proposing without asking, no concrete numbers, claiming anything can be automated, no references and creating urgency.
Four: they ask first, set limits, propose starting small and raise the exit themselves.
It means they don’t know the boundaries. An experienced partner also explains what can’t be automated.
On four headings: scope, timeline, measurement and handover. The difference is scope, not price.
It has usually left out setup, training or maintenance — and that item arrives later as an invoice.
Whether the project succeeded stays a matter of opinion. Which number will improve should be stated up front.
You may be the first attempt. That isn’t always bad but it should be reflected in price and scope — and you should decide knowing it.