Gains:
- Making the right make/buy decision by building core differentiating capabilities and purchasing standard capabilities
- Ability to evaluate suppliers with data security, compliance, ease of exit and total cost criteria beyond price
- Protect against vendor lock-in and prove value with a pre-major contract pilot
Most organizations outsource much of their AI talent: off-the-shelf models, platforms, software services, consulting. These purchasing decisions create large budgets, long contracts and permanent dependencies; When done wrong, it is expensive to come back. In this unit, you will learn how a senior executive can intelligently purchase AI solutions, decide whether to build or outsource a talent, and how to manage suppliers. The aim is not to choose the cheapest or most flashy tool, but the one that best suits the organization's needs, risks and strategy.
Make or buy?
The key decision: should you develop an AI talent internally or outsource it? The rule is simple but frequently violated: make core capabilities unique to the organization that create competitive advantage; Buy standard, common, non-differentiating skills. A predictive model with your own customer data that your competitor cannot make is the core; A text summarization tool is standard and available for purchase.
criterion
build
buy
differentiation
high, core
low, standard
speed
slow
fast
Cost
High upfront investment
Predictable subscription
control
full
limited
addiction
low
high
Hint: "Can I differentiate myself from my opponent without this ability?" ask. If the answer is "yes", buy; If it's "no, that's my difference", seriously consider doing it. Outsourcing core talent means hiring competitive power in the long run.
Supplier selection criteria
Choosing an AI vendor solely on price and features is dangerous. A manager should also ask about these criteria:
- Data security and privacy: Where is our data processed, stored, used in education?
- Compliance: Does the supplier support KVKK/EU AI Law requirements?
- Transparency: How does the model work, what are its limits, has it been independently audited?
- Integration: How easily does it connect to our existing systems?
- Ease of exit: Can we move our data and workflow if we leave?
- Sustainability: Is the supplier financially sound, will it disappear tomorrow?
- Total cost of ownership: License + integration + training + supervision (Unit 5).
Risk of supplier lock-in
A critical concept: vendor lock-in is becoming so dependent on a provider that leaving is practically impossible (due to high cost, data portability, integration complexity). This leaves you open to price increases, poor service, and loss of innovation. To mitigate this: put data portability into the contract, opt for open standards, work with multiple vendors if possible, and prepare an exit plan.
Step by step: purchasing process
1. Define the need clearly. Which business problem, which metric? (Unit 4-5)
2. Make your make/buy decision. Core or standard?
3. Evaluate suppliers with a list of criteria. Besides price, security, compliance, exit, TCO.
4. Try it with the pilot. Prove it with a small, measurable trial before the big contract.
5. Sign the contract with the guards. Data ownership, privacy, exit, service level (SLA) clauses.
three mini cases
Case 1 — Core ejection error. A logistics company entrusted route optimization, its greatest competitive advantage, entirely to an external supplier. Two years later the supplier doubled the price; The company was locked in, it had no data and models of its own. He eventually had to install it indoors with an expensive switch. Core talent should not have been purchased.
Case 2 — The privacy question changed purchasing. Before a healthcare company buys an AI tool, it asks "is our data used in training?" he asked. The answer was "yes, by default"; this was unacceptable for patient data. The company turned to an alternative that does not use the data in education and processes it domestically. One right question prevented a serious compliance violation.
Case 3 — Pilot's savings. A retailer was about to sign a 3-year, 4 million TL contract for a fancy artificial intelligence platform. First he did an 8-week paid pilot; The platform did not fit into the actual workflow of sales teams, adoption remained low. Major contract canceled, millions of pounds and years of deadlock avoided. Pilot was the cheapest insurance.
Four copyable templates
1) Make/buy analysis:
Your role: technology purchasing consultant. Analyze the make-or-buy decision for the following AI talent: [talent]. Evaluate in terms of differentiation, speed, cost, control and dependency; Question whether this is core or standard for us and give a suggestion.
2) Supplier evaluation criteria list:
Produce a checklist to evaluate a supplier for purchasing the following AI solution: data security/privacy, compliance (KVKK/EU AI Law), transparency, integration, ease of exit, sustainability, total cost. For each criterion, write a clear question to ask the supplier. Solution: [text]
3) Contract protection clauses:
List the protection clauses I would want in an AI vendor contract: data ownership, non-use of data in education, confidentiality, service level (SLA), opt-out and data portability, liability. For each clause, explain in one sentence why it is important.
4) Pilot design:
Design a 6-8 week pilot to test this AI tool before a big contract: what criteria, what user group, what is the success threshold, what decision (sign/cancel) will I make based on what outcome? Tool: [text]
Weak prompt / Strong prompt
Weak: “Which AI tool should I buy?”
Result: Superficial answer without context, ignoring security and dependency, possibly suggesting "what's popular".
Güçlü: "We are a healthcare company working with patient data, we are subject to KVKK. We are evaluating an artificial intelligence solution that will summarize clinical notes. Analyze whether this is core or standard for us; if we are going to purchase, determine the critical questions I will ask the suppliers about data security, use of data in education, domestic processing, ease of exit and total cost; also design a pilot before the big contract."
Result: Sensitive to context, regulatory and risk; make/buy, a purchasing approach that considers criteria and pilot together.
Common mistakes
- Buying core talent. Renting out your competitive advantage creates expensive dependency in the long run.
- Just looking at price and features. A cheap vehicle becomes expensive if safety, compliance, egress and total cost are overlooked.
- Not asking how the data is used. “Is our data used in education?” question captures most compliance risks from the start.
- Major contract without pilots. Being locked in for years without proving it with a small trial is a big risk.
- Signing with no exit plan. A supplier you can't part with leaves you vulnerable on price and service.
Caution: AI can produce a purchasing analysis or list of supplier criteria; But contract terms, data security commitments and legal clauses must be reviewed by legal and security teams. It is also the manager's responsibility to independently verify (pilot, reference, audit) a supplier's claims about its product; Marketing promise is not evidence.
In summary
Most AI talent is purchased externally, and these decisions are large, permanent, and costly. Basic principle: make core, differentiating capabilities; buy standard ones. Suppliers are attracted not only by price and features; Evaluate with data security, compliance, transparency, integration, ease of exit, sustainability and total cost. Protect against vendor lock-in with data portability, open standards, and an exit plan. Be sure to prove it with a pilot before the big contract. In the next and final unit, we will cover change management, which turns this entire strategy into real adoption.
Application task
Select an AI capability and perform a make/buy analysis with template 1 to decide whether it is core or standard. Create the critical questions you will ask the supplier with the 2nd template for a solution you decide to purchase (be sure to include data security and ease of exit). Finally, describe the criteria, duration, and decision threshold for a pre-major contract pilot in one paragraph.
checklist
- [ ] I decided whether the skill is core or standard.
- [ ] I justified the make/buy decision.
- [ ] I evaluated suppliers with criteria beyond price.
- [ ] "Is our data used in education?" I asked the question.
- [ ] I evaluated ease of exit and supplier dependency.
- [ ] I designed a pilot before the big contract.
- [ ] I made a plan to have legal/security review the contract and data clauses.