Data & AI platform by Data Analytics

AI connections require more than connectivity

A question about customer profitability often touches several systems: revenue sits in your finance system, customer agreements in CRM and deliveries in ERP. A useful AI application needs more than access; it also needs relationships and clear definitions. 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 of this data chain.

The examples on this page are hypothetical situations, not customer cases.

Meet our team

At Data Analytics

Start with the question that crosses systems

First identify the business decision you want to support. Analysing delivery reliability requires order lines, promised dates and actual deliveries. Analysing customer value may also require credit notes and service costs. This substance determines which sources matter and prevents you from opening up entire systems without a clear purpose.

Example: you want to know which customer groups generate more revenue while also causing more corrective work. CRM segments must then align with financial customer numbers and recorded service activities. Discuss missing keys and inconsistent records before asking AI to formulate an explanation. A connection alone does not resolve these substantive differences.

Distinguish data ingestion from context for AI

APIs, database connections and ETL processes can make source-system data available for further processing. The appropriate route depends on access options, volumes and source restrictions, among other factors. During setup, also discuss how changes, deletions and failed transfers are handled, and which data is actually needed.

MCP has a different role: it can provide an AI client with an agreed way to access context and tools. It does not replace data ingestion, cleansing or a data model. Keep these layers separate in your design so it remains clear where data is prepared and where AI requests information.

Make availability and freshness explicit

Do not assume a connector is available because a software package is well known. Check the relevant version, licence conditions, available interfaces and required permissions. An existing connection can be a useful starting point, but suitability for your data and question must be established. Not every source is universally plug-and-play.

Also agree how current the information needs to be. Financial figures updated daily may suit a management meeting, while an operational stock question has different requirements. Make clear which refresh an answer relies on and what users should do when the agreed freshness is not achieved.

Reuse existing data without losing meaning

If you already have useful data flows or an established data model, first examine what can be reused. An additional route directly to the same source can unnecessarily produce different outcomes. Reuse still requires checks: a dataset for monthly reporting is not automatically suitable for questions about individual orders.

Document how entities and amounts relate to one another. Consider customers with multiple account numbers, amounts in different currencies and revenue before or after adjustments. Data Analytics supports connecting and structuring this data in ENABLE, enabling dashboards and AI applications to work from an aligned data foundation.

Manage the full route from source to answer

A change in a source system can affect several applications. Name who receives change notifications, who performs data checks and who decides whether an application should temporarily wait. Include access credential management and coordination with your software supplier in the working arrangements.

AI also requires a deliberate choice of processing route. A local LLM differs from an external AI service receiving information through MCP. External processing may occur depending on model choice and setup. Discuss what context the application may receive rather than treating source connectivity as unrestricted permission for AI use.

Frequently asked questions

Can you connect every ERP or CRM package?

That must be assessed for each package and setup. Available interfaces, permissions, versions and data structures determine the options. Data Analytics first discusses your sources and information needs before agreeing on a specific connection.

Does MCP replace an API or ETL process?

No. APIs and ETL can form part of retrieving and preparing data. MCP focuses on making context and tools available to an AI client. Both can have distinct roles within the same architecture.

Are answers always based on current data?

Freshness depends on the source and the agreed refresh schedule. Determine the acceptable delay for each application and how users can check the period and refresh status used.

Do you need to replace existing integrations?

Not necessarily. Existing integrations may be useful when their data, definitions and management fit the intended use. Assessing what is already in place first helps you make focused decisions about additions or changes.

Discuss your question with Data Analytics

Which sources, users and checks does your application need? Discuss your question in an ENABLE demo.