Almost every System Integrator claims domain knowledge as a strategic advantage: “We understand banking.” “We know healthcare.” “We have decades of experience in retail, manufacturing or insurance.”
For a long time, this was considered a genuine moat. A consultant who already understood the terminology, processes, regulations and common exceptions of an industry could become productive faster and reduce delivery risk. But is domain knowledge still a moat when AI can compress months of learning into a few days?
1. Learning Was Always Possible. AI Changed the Speed.
Consultants could always research an unfamiliar industry.
They could read books, study process documents, interview subject matter experts, attend training programmes and learn through projects. AI did not suddenly make learning possible.
It changed the speed, accessibility and quality of the learning experience.
When I was in college, we were given assignments on internal combustion engines, boilers and other mechanical systems.
Understanding one of them could take weeks. We had to find the right books in the library, get them issued, study complicated two-dimensional diagrams and then try to reconstruct the machine in our minds. For a 16-year-old, imagining a three-dimensional engine from a cross-sectional drawing was not easy. After a month of effort, many of us probably understood parts of it incorrectly.
Today, a fifth-grade student can watch a three-dimensional animation of the same engine. The student can pause it, rotate it, separate every component and see exactly how fuel, air, pressure and motion interact.
Within minutes, the child may develop a clearer conceptual understanding than we developed after several weeks.
AI is bringing the same change to domain learning.
A consultant entering insurance can ask AI to explain the complete claims lifecycle, create a process map, compare health and motor claims, describe common exceptions, identify the systems involved and generate questions for a discovery workshop.
A process that may once have taken months of meetings, documents and project exposure to understand at a basic level can now be explored in days.
2. Does Domain Knowledge Pass the VRIO Test?
The VRIO framework provides a useful way to test whether a capability can create sustained competitive advantage. A resource must be valuable, rare, difficult to imitate and supported by the organisation using it.
The framework grew from Jay Barney’s resource-based view of competitive advantage. Barney’s foundational 1991 paper.
Valuable
Domain familiarity speeds discovery, prevents basic mistakes, and accelerates project initiation.
Rare
Standard terminology, typical workflows, and published regulations are now accessible through AI.
Inimitable
Generative AI and reasoning models make baseline domain expertise easier for competitors to acquire.
Organised
Firm processes capture knowledge, but without rarity or inimitability, it achieves parity—not moat.
Domain knowledge is certainly valuable. But is it still rare? Is it difficult to imitate?
If an SI’s domain expertise consists primarily of terminology, standard process maps, generic user stories, reference architectures and publicly available regulatory knowledge, AI makes much of it easier for competitors to acquire.
It may still be valuable, but it is no longer sufficiently rare or difficult to imitate. In VRIO terms, it may produce competitive parity rather than sustained advantage.
This is an important distinction. Something can be essential without being differentiating.
The Electricity Analogy: Essential vs. Differentiating
Electricity is essential to a technology company, but access to electricity is not its moat. In the same way, domain knowledge may become essential for every SI without remaining a credible point of differentiation.
3. What Happens to Porter’s Five Forces?
Porter’s Five Forces offers another useful perspective without requiring us to turn this into a classroom exercise. The framework examines rivalry, buyer power, supplier power, substitutes and new entrants within an industry. Harvard’s Institute for Strategy and Competitiveness explains the Five Forces framework here.
AI affects several of these forces in the SI industry:
Lower Barriers into New Domains
The barrier to entering a new domain becomes lower because a consulting company can build baseline knowledge much faster.
Clients Independently Research
Buyer power increases because clients can independently research solutions, challenge assumptions and assess proposals.
Internal Teams & AI Platforms
The threat of substitutes grows because internal teams, specialist firms and AI-enabled platforms can perform work that previously required a large consulting team.
Dozens with Identical Knowledge
Rivalry increases because dozens of SIs can acquire similar knowledge and present similar capabilities.
When everyone can learn the standard process faster, saying “we understand your industry” becomes less persuasive.
The Natural Buyer Question
The buyer will naturally ask: “What can you do with that understanding that others cannot?”
4. Domain Knowledge Is Not One Thing
The problem may be that we use the term “domain knowledge” for several very different capabilities.
| Type of domain knowledge | Example | AI’s likely impact |
|---|---|---|
| Vocabulary | Understanding industry terminology | Highly accessible |
| Standard processes | Knowing how claims or underwriting generally works | Increasingly accessible |
| Published regulations | Finding and summarising applicable rules | Faster, but requires validation |
| Enterprise context | Knowing how one company actually operates | Difficult to acquire externally |
| Exception knowledge | Understanding what happens outside the standard workflow | Partly tacit |
| Consequence-aware judgment | Knowing what could go wrong after a decision | Built through experience |
| Stakeholder trust | Having credibility when making a difficult recommendation | Not downloadable |
AI is rapidly commoditising the first three categories.
The remaining categories may become more valuable precisely because the basic knowledge is becoming abundant.
The Fraud Exception
For example, AI can explain the standard loan approval process. It may even generate a credible process diagram and list the usual controls. But a genuine banking expert may know that one apparently unnecessary approval was introduced after a particular fraud incident. That person may understand which exception looks harmless in the data but creates significant regulatory exposure.
The Scheduling Collision
AI can describe how hospitals normally schedule patients. A healthcare practitioner may know why a theoretically efficient scheduling model could fail when emergency cases, clinical priorities, doctor behaviour and patient anxiety collide.
Replenishment Nuances
AI can describe retail replenishment. An experienced operator may understand why the same inventory rule cannot be applied to a supermarket, a luxury retailer and a seasonal fashion business.
The difference is no longer access to the process. It is the ability to understand context, exceptions and consequences.
5. There Is an Even Bigger Question: Process Redesign
Most discussions assume that the purpose of domain knowledge is to implement an existing industry process correctly.
But what if the process itself is due for redesign?
Many enterprise processes were created when information was difficult to access, computing was expensive and departments operated through separate systems. Manual reviews, sequential approvals and reconciliation activities were necessary because technology could not observe, reason or act continuously.
Now that technology, data and computing capacity have changed, should we continue to automate these processes as they are?
Or should we first ask why they exist?
An SI may have implemented the same workflow fifty times. That experience can reduce risk on the fifty-first implementation. But it can also make the SI deeply invested in preserving the workflow.
Experience teaches us what normally works. Transformation requires us to recognise when “normal” is no longer relevant.
6. The Moat Is Shifting from Knowledge to Judgment
AI adoption is already widespread. Stanford’s 2025 AI Index reported that 78% of surveyed organisations used AI in 2024, compared with 55% in 2023. Stanford AI Index 2025.
The World Economic Forum also reports that employers expect 39% of workers’ existing skill sets to change or become outdated by 2030. Future of Jobs Report 2025.
These numbers do not prove that domain expertise is disappearing. They indicate that knowledge-based advantages cannot be assumed to remain static when the speed of acquiring and applying knowledge is changing.
Perhaps the new SI moat is not domain knowledge itself, but a combination of:
Domain Judgment
Evaluating subtle trade-offs and knowing what could go wrong after a decision.
Enterprise-Specific Context
Understanding how the individual organization works, its legacy nuances, and unwritten operational realities.
Process Redesign Capability
Challenging historical workflows rather than merely automating existing inefficiencies.
AI & Engineering Depth
Robust engineering rigor to build, evaluate, and scale production systems natively.
Trust & Accountability
Standing behind recommendations and taking responsibility for production outcomes.
Evidence from Successful Execution
Real delivered systems and proven track records that slides cannot replicate.
Knowing how an insurance claim currently works is domain knowledge. Knowing which controls must remain, which activities can be eliminated and how an AI-native claims process should operate is strategic domain capability.
Knowing a banking process is useful. Knowing why it exists, where it fails and how it can be safely redesigned is harder to imitate.
7. The Domain Expert Must Also Evolve: From Translator to Challenger
The traditional domain expert often acted as a translator.
The business explained its problem. The domain expert converted it into requirements. The technology team converted those requirements into a system.
The future domain expert may need to become a challenger.
That person must distinguish between:
Most enterprises describe all five as “the process.”
The ability to separate them is far more valuable than remembering the process flow.
AI engineers without domain judgment may redesign something they do not fully understand. Traditional domain experts without technology awareness may protect processes that no longer deserve to exist.
The strategic advantage lies in combining both.
8. So, Is Domain Knowledge Still a Moat?
I am not convinced that generic domain knowledge is a moat anymore.
It remains necessary. It improves conversations, accelerates discovery and reduces mistakes. But AI is making standard industry knowledge faster and cheaper to acquire.
What previously took years may take months. What took months may take days. And what took several days of reading may now take one well-structured conversation with AI, followed by proper validation.
The real moat is moving towards contextual knowledge, lived experience, critical judgment, relationships, accountability and the ability to redesign processes rather than merely reproduce them.
Perhaps SIs should stop saying:
The Stronger Proposition:
«“We understand your industry well enough to know what must be protected, what should be challenged and what needs to be reinvented.”»AI has not made domain knowledge irrelevant.
It has made generic domain knowledge less defensible as a strategic advantage.
And that raises a difficult question for every System Integrator:
If domain knowledge is no longer your moat, what is?
The future belongs to partners who blend deep architectural engineering with the courage and judgment to reinvent legacy enterprise operating models.
