Consultancy evaluation · Step 3 of 5
Understand the AI-era advantage
See why approved definitions, governed architecture and reusable design context become more important as AI generates a larger share of technical implementation work.
The strategic change
AI can accelerate implementation. It does not automatically protect meaning.
Coding assistants and business-user agents can increasingly generate schemas, SQL, pipelines, semantic models and documentation. This can reduce the time required to produce technical assets.
The remaining challenge is deciding what those assets should represent, which definitions they should use, how they should fit together and whether they follow the approved architecture.
AnalyticsCreator gives teams a structured design foundation from which dependent technical assets can be generated. This helps ensure that faster implementation does not create inconsistent definitions, duplicated logic or disconnected technical layers.
The design context
Give people and AI the same approved foundation
Microsoft and other technology providers supply the agents, coding tools and execution environments. AnalyticsCreator can provide the approved design context from which engineers, consultants and AI-assisted workflows build.
Agreed business meaning
Define what business concepts, measures, entities and relationships mean before they are translated into schemas, SQL or analytical outputs.
Approved delivery structure
Establish the target layers, models, historisation approach, transformation patterns and deployment structure that the implementation should follow.
Reusable engineering logic
Capture transformation, loading, naming and modelling rules in structured metadata rather than relying on isolated prompts or undocumented code.
Traceable dependencies
Maintain the relationships between source metadata, transformations, warehouse structures, semantic models and documentation.
Consistent implementation
Give different consultants, engineering teams and AI tools a shared basis for producing compatible technical assets.
Controlled regeneration
Update the design and regenerate dependent assets rather than asking separate tools or agents to reinterpret the change independently.
A simple definition
What does “ontology” mean in this context?
An ontology is a structured description of the important business concepts in an organisation and the relationships between them.
For example, it can define what a customer, order, product, contract or revenue measure means, how those concepts relate to one another, and which rules apply when they are implemented in a data solution.
Business meaning
- What the organisation calls a customer
- How revenue is defined and calculated
- Which relationships exist between business concepts
- Which rules and classifications are approved
- Which definitions are shared across reports and systems
Technical implementation
- Tables, columns and relationships
- SQL transformations and loading logic
- Warehouse and dimensional models
- Semantic models and measures
- Pipelines, deployment assets and documentation
When business concepts are defined separately from a specific schema, tool or codebase, they can be reused across technical implementations and survive future platform changes.
The risk without shared context
Faster generation can also produce faster inconsistency
AI tools can generate technically valid outputs from the information they are given. Problems arise when different users, tools or agents work from different assumptions.
Definition drift
Different agents or project teams may implement the same business concept differently because no approved definition has been supplied.
Architectural drift
Generated assets may work individually while failing to follow the intended layering, historisation, naming or deployment approach.
Logic duplication
Business rules may be recreated independently in SQL, pipelines, semantic models and reports rather than being governed from one design foundation.
Undocumented decisions
Prompts and generated code may contain important assumptions that are not preserved as reusable, reviewable project knowledge.
Difficult validation
Reviewers must assess each generated asset separately when there is no approved model against which the output can be checked.
Tool-dependent knowledge
Important definitions may become embedded in a particular agent, prompt library, schema or vendor implementation rather than remaining portable.
The consultancy opportunity
Move from producing code to governing what gets built
As implementation becomes easier to generate, the differentiated value of a consultancy increasingly lies in the quality of its architecture, business understanding, delivery standards and ability to control change.
Lead the design decisions
Help the customer define the architecture, business meaning and delivery standards that both engineers and AI-assisted tools should follow.
Preserve delivery knowledge
Capture proven modelling and implementation patterns in a structured form that can be reused across projects without reducing every engagement to a fixed template.
Validate outputs against an approved design
Give technical reviewers a defined target against which generated assets can be assessed, rather than reviewing isolated code without broader context.
Align delivery teams and tools
Help coding agents, business-user agents and partner delivery teams work from the same approved definitions and architecture.
Keep knowledge portable
Separate the approved design context from a single AI model, prompt interface or production runtime so that the customer retains control of the solution.
Govern future change
Use the maintained design model as the basis for extensions, migrations, new use cases and AI-assisted development over the life of the customer relationship.
Practical operating model
How the roles fit together
Customer and consultancy define
Agree the business concepts, target architecture, delivery standards and customer-specific rules.
AnalyticsCreator structures
Capture the approved design, relationships, rules and dependencies in a reusable metadata model.
Tools and agents implement
Generate or assist with the technical assets required for the selected Microsoft environment.
The consultancy validates
Review the implementation against the approved design, test the solution and manage delivery into the customer environment.
The consultancy remains responsible for turning customer requirements into the correct architecture, governing the design and validating the resulting solution.
Evaluation questions
Test whether this matters in your delivery business
- Are your consultants already using AI tools to generate SQL, scripts, models or documentation?
- Do different project teams interpret the same business concepts or architectural standards differently?
- Is important delivery knowledge stored in code, prompts, documents and individual experience rather than one governed model?
- Could you clearly show an AI tool which definitions and architectural patterns are approved?
- Can generated outputs be validated against an agreed design before they are deployed?
- Do business definitions survive when customers change databases, integration tools or analytics technologies?
- Could a maintained design context support several engineering teams, agents and delivery partners at the same time?
- Is the consultancy able to retain its architecture and governance role as implementation becomes easier to generate?
The question is not whether AI can generate code. It is whether your consultancy has a governed design foundation that ensures AI, Microsoft tools and engineering teams build the right solution consistently.
