inSTREAM™ Learning & Intelligence

Celaton inSTREAM™ learns autonomously to perform tasks that normally require human intervention and intelligence such as perception and decision-making to provide a new level of artificial intelligence to the automated process.

Problem

The inbound information streams that organisations receive everyday are both unpredictable and unstructured. They contain valuable and irrelevant information alike, yet both demand valuable resources to process it and get it where it needs to be.

Traditional technology can deployed to capture and extract data from documents as long as the document is received in a structured format. However, even the smallest adjustments to a document such as an extra line item on an invoice or a slightly mis-aligned scanner can create errors which need to be resolved by an operator. More important, these operators will continue to be relied on to handle the same errors unless a suitable template can be programmed to deal with potential variations.

Imagine your concern if you could not rely on staff because they were consistently unable to understand a document simply because the data they were looking for was even a few millimetres away from where it should be on the page, and yet traditional technology continues to be unreliable in the same way.

Solution

Unlike traditional technology, humans learn by experience. Their accumulated knowledge is what continues to make them essential to the handling of a transaction. However, to minimise or even eliminate the need for human intervention, this knowledge must not be confined within the individual operator but acquired by the system to benefit the handling of all transactions.

Essential to the efficiency of inSTREAM™ is its ability to learn. Whilst human intervention is initially required to assist or simply agree with the processing of a transaction, every action taken by an operator to assist (or agree) with a transaction teaches inSTREAM™. This accumulated knowledge ensures that the need for future manual resolution of the same transaction is eliminated. This artificial intelligence significantly drives down the need for human intervention and enables efficiencies that were previously out of reach.

Results

Transactions that require no human intervention enable organisations to realise significant new levels of efficiency, visibility and compliance. inSTREAM™ is delivered as a service with an average deployment time of less than 4 weeks and the ability to scale rapidly to handle increasing transactions volumes that include anything from a till receipt to a complex contract from anywhere in the world.