Data & AI platform by Data Analytics
AI governance starts with clear responsibility
When employees use AI for analysis, answers or applications, questions arise that go beyond choosing a model. Who decides which revenue definition applies? What information may a user access? And who reviews a proposed action? AI governance makes these agreements concrete. ENABLE, the Data & AI platform from Data Analytics, connects and structures reliable data for controlled dashboards, analysis, and developing and sharing apps and AI. Data Analytics supports setup and management.
The examples on this page are hypothetical situations, not customer cases.
Meet our team
01
Assign ownership to both data and applications
Start with the business processes where you want to use AI. A finance lead may own reporting definitions, while a process owner determines what support is appropriate for order processing. Record who may approve changes, who handles questions and who decides when an application should temporarily not be used.
Also distinguish between business ownership and technical management. An administrator can maintain a connection without deciding what a margin definition means. Data Analytics helps translate these responsibilities into a workable setup, so decisions about data, models and usage do not fall between different teams.
02
Link access roles to recognisable tasks
Determine which data each user group needs for its work. A sales team may need customer revenue but not individual salary details. Describe access not only by source, but also by data type, application and action. Agree how you handle new starters, role changes and employees leaving.
Example: a regional manager may analyse performance for their own branches, while the executive team discusses consolidated figures. During setup, examine how this scope carries through to dashboards and AI applications. A written agreement is insufficient if an underlying data route makes more information available than the task requires.
03
Document definitions alongside model choices
An AI answer is difficult to assess when terms remain unclear. Define revenue, active customer and available stock, for example, including the period, exceptions and responsible owner. These agreements give users a shared reference point and reveal when two applications answer the same question using different assumptions.
Then choose a model based on the task and the conditions for data processing. A local LLM differs from an external AI service receiving context through MCP. With external models, processing may take place outside your environment, depending on your choice and setup. Document the reasoning and usage boundaries.
04
Review proposals before anything changes
Clearly distinguish between retrieving information, drafting a proposal and executing a change. A summary of outstanding receivables requires a different review from changing a payment arrangement. Decide in advance which actions require an employee to check the information used, the reasoning and the proposed consequences.
Example: AI drafts an explanation for an unusual cost trend. The budget owner checks the period, entries and explanation before the text appears in a management report. If you later add functions that execute actions, separately assess what explicit permission is needed and how errors can be corrected.
05
Manage AI products throughout their use
Governance does not end when an application becomes available. Sources change, definitions evolve and users encounter new situations. For each AI product, record its purpose, who is responsible and when you will review its operation. Include dependencies on connections, data models and selected AI services.
Data Analytics supports you in setting up and managing these relationships around ENABLE. Start with a clearly scoped application and discuss real user questions during reviews. This makes governance part of day-to-day management without treating every exploratory analysis as though it immediately executes a business decision.
Frequently asked questions
Is AI governance mainly an IT responsibility?
No. IT is important for setup and management, but business owners determine definitions and acceptable use. Management sets responsibilities and decision boundaries. You need these roles together to make agreements workable.
Does every AI output need prior approval?
That depends on the use and potential consequences. Internal exploration requires different checks from publication or a data change. Define for each application when human review is required and who is responsible.
Does local AI use remove the need for governance?
No. Access, definitions, data quality and review remain important with a local model. Local processing addresses only part of the setup questions and does not, by itself, establish whether an answer is correct.
Where should you start if no agreements exist yet?
Choose one concrete application, name an owner and describe the required data and permitted actions. Use that scope to document roles, model choice and review before expanding usage further.
Discuss your question with Data Analytics
Which sources, users and checks does your application need? Discuss your question in an ENABLE demo.
Meet our team