What Does an AI Consultant Do? What Companies Should Expect

That is a broader responsibility than recommending AI tools.

Effective AI consulting connects business priorities, processes, data, technology, governance and people.

The objective is not to maximise the amount of AI used across the organisation.

It is to make better investment decisions about where AI belongs.

What does an AI consultant actually do?

An AI consultant should help leadership answer a sequence of practical questions.

• Where could AI materially improve the business?
• Which opportunities are worth pursuing?
• Do we have the required data and systems?
• What should we build, buy or integrate?
• What risks and governance requirements need to be addressed?
• How will the solution fit into existing operations?
• And how will we know whether it worked?

Typical responsibilities therefore include:

Business and process assessment

• AI opportunity identification
• use-case prioritisation
• data and technology readiness assessment
• business-case development
• AI strategy
• AI governance
• architecture and technology decisions
• implementation planning
• delivery and integration
• change and adoption
• performance measurement

Should AI consulting start with technology?

Usually not.

Starting with a model or platform can lead organisations to search for problems that justify the technology.

A stronger approach begins with operational reality.

Where is expertise difficult to access?

Where do employees spend substantial time on repetitive knowledge work?

Where are manual handovers slowing down processes?

Which activities require employees to repeatedly search across documents or systems?

Where could automation increase capacity?

Once the problem and desired outcome are clear, technology can be evaluated objectively.

The answer might be generative AI, an intelligent automation, a private enterprise AI environment, an integrated AI assistant or a custom application.

In some situations, the right answer may not involve AI at all.

A good consultant should be comfortable making that recommendation.

How should AI use cases be prioritised?

Most organisations can identify dozens of possible AI applications.

The challenge is deciding what deserves investment.

A strong prioritisation model considers at least five dimensions.

Business value

What measurable improvement could the use case create?

Feasibility

Are the necessary data, systems and capabilities available?

Implementation effort

How complex are the integration and organisational dependencies?

Risk

What data is involved, what could go wrong and where is human oversight required?

Adoption

Will the solution fit naturally into the way people actually work?

This prevents organisations from selecting projects simply because they are technologically impressive.

A relatively straightforward use case that removes substantial repetitive work may generate more value than a sophisticated AI initiative with no clear operational owner.

What should an AI consulting engagement deliver?

The outcome should be more than workshops and a strategy presentation.

Business

Clear understanding of the problems worth solving

Opportunities

Prioritised AI use cases

Readiness

Assessment of data, systems and organisational capabilities

Business case

Expected value, effort and success criteria

Strategy

Decisions on where and how AI should be adopted

Governance

Responsibilities, policies and controls

Technolog

Appropriate architecture and technology choices

Roadmap Priorities,

dependencies and implementation sequence

Delivery

Path from concept to an operational solution

Measurement

Defined indicators for business impact

The real output of AI consulting is therefore decision clarity.

Leadership should know where to invest, where not to invest and what needs to happen next.

Why does AI governance matter?

AI governance should develop alongside the strategy rather than after implementation.

Organisations need clarity about:

• approved AI systems
• permitted data
• access rights
• accountability
• human oversight
• testing and quality
• documentation
• risk management

These questions become particularly important when AI interacts with confidential corporate information, personal data or business-critical processes.

The Swiss regulatory environment is also evolving. Switzerland currently has no overarching AI-specific legislation, while the federal government is preparing proposals for new rules on AI use as part of work scheduled through the end of 2026.

Companies operating across multiple jurisdictions may also need to consider requirements beyond Switzerland.

Governance should therefore be part of architecture and implementation decisions from the start.

Should consultants remain involved through implementation? There is significant value in connecting strategy with delivery.

Many AI initiatives fail to create value not because the initial concept was poor, but because operational realities were underestimated.

Production implementation may involve:

• integration with existing systems
• identity and access management
• data preparation
• security controls
• testing and evaluation
• process redesign
• user experience
• employee training
• ongoing monitoring

A strategy developed without understanding these dependencies can quickly become unrealistic.

Consultants who understand delivery can make better strategic decisions.

Delivery teams that understand the business case can build better solutions.

What role should employees play? AI transformation is also organisational transformation.

Employees often know where the strongest opportunities exist because they understand the real process, including exceptions that may not appear in formal documentation.

Their involvement helps identify practical use cases and improves adoption.

This matters because technical deployment is not the same as business adoption.

A solution that functions correctly but is rarely used does not constitute a successful AI transformation.

How should AI success be measured? Success criteria should be defined before implementation.

Depending on the use case, they could include:

• reduced processing time
• lower administrative effort
• fewer manual steps
• faster information retrieval
• increased employee capacity
• reduced error rates
• improved response times
• improved service quality
• user adoption

The organisation also needs a baseline.

Without understanding current performance, it is difficult to demonstrate whether AI created an improvement.

AI initiatives should therefore be measured as business investments, not simply as technology deployments.

What are the warning signs when choosing an AI consultant?

Be cautious when the conversation begins with a product rather than your business problem.

Other warning signs include:

• every problem appears to require AI
• no clear method for prioritising use cases
• ROI is discussed only in general terms
• data and security questions are postponed
• governance is treated as someone else's responsibility
• implementation is outside the conversation
• success cannot be measured

A credible consultant should also be willing to recommend not pursuing a particular use case.

Independent judgement is part of the value of consulting.

What should you know by the end of the engagement?

You should be able to answer five questions clearly:

Where should we use AI? Why are those opportunities worth pursuing? What technology, data and organisational changes are required? How will we implement and govern the solutions? How will we measure the value created? If those questions remain unanswered, the organisation may have received AI advice, but it does not yet have an actionable AI strategy.

From AI opportunity to implementation inPositiv helps organisations connect AI strategy with practical delivery - from identifying and prioritising opportunities to governance, architecture, implementation and adoption.

Explore our AI consulting services in Switzerland