All articles →

Have ERP, CRM or a webshop but still work manually? AI can connect what your systems don't cover

Controlling and reporting

You have ERP, CRM and a webshop, yet employees still transfer information between them. A customer enquiry arrives by email, someone checks the details, opens a CRM record, prepares a document and manually tells the next person what to do. The software is there, but people still hold the process together.

The biggest opportunity for improvement is often between existing systems. Connecting data, business rules and AI processing can reduce work repeated for every enquiry, document or new product.

Why does more software not automatically mean less manual work?

Each system covers part of the business. ERP manages business records and transactions, CRM tracks customer relationships and the webshop supports sales. Problems arise when information from one process is not ready for the next, or when a clear transfer between systems is missing.

An employee becomes the connection: reading documents, interpreting content, copying information, filling gaps and checking that everything has reached the right place. This creates waiting, duplicate entry and greater dependence on individuals.

Where does integration help, and where does AI add value?

When information is already structured and the transfer rule is clear, conventional integration or automation may be enough. Moving a defined field from one system to another is an example.

AI can be useful when the input is less structured: a customer email, supplier PDF, scanned invoice or product description assembled from different sources. It can help recognise content, classify it and prepare a structured proposal for the next step.

A solution may therefore combine approaches. Integration moves information, business rules define permitted actions and AI processes content that needs interpretation. Important decisions and exceptions remain subject to agreed checks.

Four areas to consider for a first project

New product preparation. Information from catalogues, images and identifiers becomes attributes, descriptions and records for an existing system. The NOS ItemAI example shows how this can become a more consistent workflow.

Document and invoice processing. Instead of reading and copying manually, a document is processed, information is extracted and checked against rules, and exceptions go to a responsible employee. DokAI and UraAI are presented for these areas on the NOS AI platform.

Outstanding receivables. An account statement can start a workflow that separates items by customer, applies rules and prepares communications. Our NOS IOSAI case study describes this application.

Classifying business enquiries. A possible project is turning enquiry content into structured information and preparing the next step for a sales team. Sources, mandatory fields and checks need to be defined before development. This is an example for assessment, not a claim that every CRM already has a ready-made NOS integration.

How can you start without replacing your systems?

Choose one frequent process with a clear beginning and end. Describe its input, what an employee does and what result the next person needs.

We then assess data sources, available APIs or file exchange options, permissions and business rules. A limited trial scope with representative examples follows. Expansion makes sense when output quality and business value have been demonstrated.

Define from the outset what happens when information is missing, a system is unavailable or a result needs human review. Useful automation includes a clear approach to exceptions.

How do you know you have chosen the right process?

A good candidate has sufficient volume, repeatable steps and a cost you can track. Measure processing time, manual transfers, corrections and time to completion.

Does every process need AI? No. If a task can be handled reliably through rules or integration, that may be a simpler choice.

Does connecting systems mean everything runs without oversight? No. The level of automation and required approvals depend on the process and the consequences of an error.

Show us a process still held together by spreadsheets and email

You do not need to know which AI solution you need in advance. Describe where the same information is handled repeatedly, where work waits and what employees still check manually.

Explore our approach to business process automation and share an example from your organisation. Together, we can assess where connecting systems and AI processing could deliver a useful result.

Have a process you
would you like to improve?

Tell us what is slowing your business down today. Together we will find opportunities to simplify, connect systems or automate.