A look at current approaches and market developments

11. January 2021

Authors: Tina Messner, Intern & Daniel Wittwer, Senior Consultant

 

What is input management?

Companies receive all kinds of information from numerous input channels – such as physical mail, emails and Chatbot messages. Input management focuses on methods, concepts and techniques that can classify and evaluate these items and digitally capture their business-relevant data so that it is available for further processing or document management.

 

Current situation

Technological advances are being made in digitalisation and digital transformation because of artificial intelligence (AI) and machine learning (ML).  Established providers of input management solutions are expanding to take account of this technology, and new offerings and products are entering the market.

New approaches such as using modern AI technology and cloud-based operating models are increasingly allowing young companies to establish themselves in the document classification and data extraction markets. As a result, the number of solutions and offerings are increasing.

Since our foundation in 2009 Inacta has been active in the field of information management – input management is one part of this. How have the approaches, processes and solutions used in input management changed? In order to perfect and expand our consulting expertise, we have taken a closer look at this market.

 

AI solutions need time

The benefit of the automated classification of documents and related data extraction (for example invoice numbers or addresses) is easy to see: They optimise processes and the potential to automate resulting in increased efficiency and reduced costs.

To achieve these improvements, many approaches use AI, although this is not always the case, even in respect of AI solutions already on the market. That is because AI - just like traditional IM solutions - must be 'trained’ to realise its full potential. This type of ‘training’ of AI intelligence typically involves speech analysis and images of a scanned document. The coordination and resources needed to train AI intelligence in this way should not be underestimated.

Different solutions have different focal points when it comes to classification and data extraction. For example, one solution might focus on the classification of only a small number of test documents; another might focus on the validation of extracted data compared to master data.

To avoid disappointment, it is crucial that there are no false expectations of AI-based solutions. To avoid this an in-depth analysis of existing processes and the extent to which customers can benefit is needed.

 

 

As a rule, different solution providers also offer, to some extent, handwriting recognition, reading of different languages and different options for the post-processing of extracted data. They also differ in their basic approach to classification and extraction: for example, one provider focuses on the intuitive and independent training of the AI solution, so that customers can stipulate what they require. Other providers focus more on the use of already trained standard documents, which minimises the training needed and shortens the time-to-market.

 

Establish precisely what is needed

It is essential to determine exactly what is required from input management – it could be a fully comprehensive input management solution allowing the integration of an e-mail inbox and downstream business process.   In that instance, what is needed is quite different compared to a focus on the extraction of data, such as invoice data used in the accounts payable process.

Available solutions usually have as standard features the support of document classification and the extraction of common data contained in enterprise resource planning (ERP) systems.  Documents routinely supported include invoices, delivery notes and purchase orders. However, solutions can also be tailor made depending on need, the industry concerned and the complexity of the documents’ structure.

These are examples of the different features and requirements that an AI-based document classification and data extraction solution might need to address.

As solutions are sourced from the cloud, all service providers focus closely on good data protection and compliance. This is usually achieved by ensuring that servers are located in either Switzerland or the EU and have ISO certification.

 

Conclusion – How Inacta can help

It is always important to carefully analyse a customer’s needs and starting position so that new technology can be used successfully. Here Inacta can provide support using our years of experience and expertise dealing with tried and tested solutions. As well as providing a project management service and analysing your requirements, we can share our market knowledge and support your project from kick-off to go-live.

 


Contact person:

Jürg

Jürg Porro

Head Business Consulting & Healthcare
E-Mail
LinkedIn