Who Should I Hire for AI — A Developer or a Consultant?
Choosing an AI consultant is the murkiest decision in this field, because everyone uses the same words: “AI solution”, “automation”, “transformation”. Same words, different work. 🔎
Short answer: first someone to write the job, then someone to build it. A project starting with tool selection goes unused — because which job it solves was never clarified.
Below: what each party does, the selection criteria, the red flags and the contract clauses. 🎯
Who does what?
Choosing an AI consultant starts by separating the roles.
Can someone in-house run it?
They can; it’s discipline rather than technical work. The sticking points are usually rule-writing and measurement — outside help accelerates those two.
What are the selection criteria?
Three filters eliminate most candidates.
What are the red flags?
Some sentences end the conversation.
Who should own it internally?
However good the outside party.
What belongs in the contract?
Four clauses protect you.
The four protective clauses
1) Data and account ownership with the business; where information goes is written. 2) A delivery list: which job, which templates, which rule set. 3) Training — who explains it to the team, for how many hours. 4) Measurement: starting times and an end-of-period comparison. With those four written, the work runs transparently; reading a quote sits in the quote article. 📜
📝 Field Notes
The most common sales method we see in this field: an impressive demo, then an annual subscription. The job shown in the demo is usually not the one costing the business most time. In our first meeting there’s no screen; there’s a one-week hour hunt. Knowing the tool is easy; knowing which job it should solve is hard. ❓
📖 Quick Glossary
Rule set: the written statement of what may be entered. Internal owner: the person running the setup inside the business. Template: a reusable text skeleton. Data ownership: information belonging to and controllable by the business.
⚡ Quick Summary
First the job writer, then the builder. 🔎 Consultant solves which job, developer solves how. A party selling a tool describes a need that suits it. A candidate not discussing data rules is out. Every project needs an internal owner.
🎯 Next Step
Put us through the same filter: our first meeting is about the job, not the tool — the first review is free: the quote page. Scope on the consulting page. ✅
Frequently Asked Questions
Sık Sorulan Sorular
A consultant determines which job gets faster, which information gets used and what stays with people. A developer solves how to build it: integration, automation, interface. These are different professions; expecting one to do the other produces disappointment. 🧭
You can, but it carries a bias: a party selling its own product tends to describe a need that suits that product. That isn’t bad faith, just a natural tendency. Our model doesn’t sell tools for exactly this reason — scope on the consulting page. ⚖️
A party opening with a tool demo aims to sell a product, not solve your job. The right party asks you questions: which task did you do how many times this week, how long does a quote take. The hour-hunt logic sits in the starting article. ❓
A party not raising what may and may never be entered in the first meeting isn’t seeing the risk. This clause can’t be added later; it’s the foundation of the setup. 🔒
Work that doesn’t measure the starting time answers “did it help?” by feel. Hours saved must be recorded up front — the method sits in the return article. 📊
“AI will take over all your processes”, “it’ll halve your staffing needs”, “give us your data and we’ll train it”. None is realistic; the third is also dangerous for data ownership. The reality of the limits sits in the jobs article. 🚩
Setup finishes, the consultant leaves. Without an owner inside, nobody fixes the first bad output and the team reverts within two weeks. The owner should be someone doing that job daily — not a manager. 🔑
Setting the rule and making the gain visible. Unless “quotes are now prepared with this template” comes from management, everyone returns to their own method. Adoption sits in the team article. 📣
Not necessarily, but clarifying which job gets solved independently is healthier. A party selling its own product naturally describes a need suited to it. The decision and the priority list should stay with you.
You can. Run the one-week hour hunt yourself, pick the most repeated job, write a template with a ready tool. Outside help brings speed and blind-spot detection. The one thing not to do is buy a subscription without knowing which job it solves.
They will, but everyone learns differently and uses it differently without rules. Short training on real work embeds both the rule and the template at once. The goal isn’t teaching a tool but building a shared method.
Source: World Economic Forum — reports