Build vs buy AI: a decision framework

Build or buy is one of the most asked questions when a business starts thinking about AI. It is also the question most consultants get wrong — usually because they make money on one side of it.

We build both. We sell off-the-shelf vertical products under XWorks Suite and we ship custom AI on the same engine. We have no financial incentive to recommend one path over the other. Here is how we actually think through it.

The seven criteria

Score each one from 1 (off-the-shelf wins) to 5 (custom wins). Add at the end.

1. Time to value

How fast do you need this in production?

• 1: "We want it next month or it does not matter."
• 5: "We can wait six months if it is the right thing."

2. Customisation depth

How specific is your workflow to your business?

• 1: Standard industry workflow. Looks like 90% of competitors.
• 5: Your competitive edge is in the unique process. Nobody else does it this way.

3. Data sensitivity

How sensitive is the data the AI touches?

• 1: Public-ish data. Customer-facing communication, marketing copy.
• 5: Highly regulated. Healthcare records, financial data, legal opinions.

4. Total cost over 24 months

Off-the-shelf is cheap upfront, expensive at scale. Custom is the opposite. At your projected scale, which wins?

• 1: Off-the-shelf monthly fees stay reasonable at your scale.
• 5: Off-the-shelf would cost more than building your own over 24 months.

5. Integration complexity

How deeply does the AI need to integrate with your existing systems?

• 1: A few standard integrations (CRM, email, payments). All exist as off-the-shelf connectors.
• 5: Deep integration into proprietary internal systems or unusual third-party APIs.

6. Team capacity

Do you have an engineering team that can manage a custom build?

• 1: No engineers. The closest we have is a tech-savvy office manager.
• 5: Strong engineering team. They want to own this.

7. Exit cost

If you walk away from the choice in two years, what is the cost of switching?

• 1: Low. Off-the-shelf vendor has reasonable export and we can switch in weeks.
• 5: High. Whatever we pick will live in our workflow for 5+ years.

Read the score

The single criterion that matters most

If you only weight one criterion, weight customisation depth. Everything else can be worked around. If your business edge depends on the AI being uniquely yours — your data, your workflow, your model — then off-the-shelf will become a ceiling within 12 months. If it does not, off-the-shelf will keep up with you longer than you expect.

When buy wins (most of the time)

Buying off-the-shelf is the right call when:

• Your industry is well-understood. The pattern of work is standard.
• You compete on execution and customer experience, not on technology.
• You are early in your AI journey and want a fast learning loop.
• You do not have the engineering team to run a custom build.

This describes most operators — small gyms, salons, clinics, restaurants, retail, finance practices. XWorks Suite exists for exactly this.

When build wins

Custom builds are the right call when:

• Your competitive edge is the AI itself. Customer experience comes from a uniquely-trained model or a uniquely-instrumented workflow.
• Your industry has no off-the-shelf product, or the off-the-shelf options have a clear ceiling.
• The integration is so deep that buying means rebuilding around the vendor anyway.
• You have or can acquire the team to operate it.

This is where our Custom AI offering lives. Same engine that powers our vertical apps, shaped to your specific workflow. Typically 4-8 weeks to a working v1.

When hybrid wins

Hybrid is underrated. It is the right call when you need most of what off-the-shelf offers, plus 1-2 things that are uniquely yours.

Example: a gym chain runs on XWFit for member CRM, billing, and attendance — but they have a proprietary AI-driven personal-training programme that nobody else has. We ship XWFit out of the box and build a custom personal-training module on top, using the same engine. Best of both.

Common pitfalls

"Let's just build it ourselves"

Said by every engineering leader, ever. The cost is almost always 3x what was estimated and the time-to-value is 2-3x slower than buying. Building should be a deliberate choice, not the default for tech-positive teams.

"This off-the-shelf product covers 95% of what we need"

Translation: the 5% gap is where 80% of the value lives. Ask whether the 5% is your competitive edge. If yes, do not buy; build that 5% or fork the 5% with a hybrid arrangement.

"We will switch later if it does not work"

AI products embed deep into workflows. Switching costs are real even when "data export" is offered. Score the exit-cost criterion honestly.

"We do not have the team to build it"

Used to be a buy-or-do-not-build signal. With AI engineering platforms (including our own), the team-size requirement for custom AI builds has dropped sharply. Get a fresh quote before assuming you cannot build.

What this means for you

• Score yourself on the seven criteria. Add the total. Honest answers only.
• If your score says buy, look at XWorks Suite first.
• If your score says build, talk to us about a custom AI build.
• If your score says hybrid, we ship both paths from the same engine — it is the right setup for that case.

Want a second opinion? Book a 30-minute call. We will tell you which side of the line we think you sit on.