Andrew Anderson 3

The art of smart

18 July 2014

Can artifical intelligence manage your information overload?

Information is streaming into businesses, all day, every day and in ever increasing volumes. It fills our office spaces and our hard drives, arriving through a growing number of channels – both real and virtual. It floods into our email inboxes and piles in through our letterboxes – mixed-up, messy and unstructured. And for many businesses it is expensive and time-consuming to deal with.

Hard copy documents are still being used by many customers, while the ease and inexpense of digital communication have increased the number of interactions. This multitude of communications channels is making it hard for businesses to keep pace with their customers and clients, and can increase errors, cycle times and costs. 

Simply throwing more people at the task is not a feasible long-term solution, so businesses must look to more intelligent processes, systems and software to improve customer service and compliance, and reduce costs and mistakes.

Equiniti has been working with the team behind one such system – inSTREAM by Celaton, which uses artificial intelligence (AI) to streamline labour intensive clerical tasks and decision-making. inSTREAM is a great example of a new generation of software changing business methods and offering new business process outsourcing opportunities.

Celaton was born in 2004, a result of a management buyout of Redrock Software from Netstore Plc and the acquisition of DG Tech. But its history goes back to 1993. Chief Executive Officer, Andrew Anderson – a former paratrooper – says their inSTREAM software as a service can transform the way enterprises handle unstructured content flowing into organisations through different channels, including financial and non-financial information received by email, fax, post and paper.

The system has the ability to learn the patterns and processes like any new employee would

“It enables scale and efficiencies that were previously out of reach and minimises the need for human intervention, ensuring that only accurate information enters the line of business systems,” says Andrew. “What I learned in more than 20 years in the IT industry is that it’s all very well delivering IT as a service but the real pressure comes from organisations’ ability to provide a customer with the right experience, including the convenience and the choice for them to be able to communicate how they want. That is often stuff that comes in by paper, and is slowly being taken over by email. All this material is really unstructured and very, very labour-intensive to process. So what Celaton really focused on was how can we apply AI to make the process more efficient? And cut out the errors and the labour – the human element – where possible.

“We spent the first six to seven years developing this technology which could understand any media in any format – creating a single channel. With inSTREAM the objective is not only to recognise it by reading it, but to understand what it means. The system is actually making decisions by recognising and extracting important information, verifying and checking it against other information and delivering it into the right business system.”

The system has the ability to learn the patterns and processes like any new employee would. To begin with it may need to flag up when it doesn’t recognise what it is dealing with and will require direction from a person, but as it processes more, it learns more, and will require fewer interventions. It relies on the ability to recognise patterns, rather than just words and phrases, and learns autonomously to match hundreds or thousands of relevant reference points to achieve the level of confidence required.


Andrew Anderson

Andrew adds: “It’s important to visualise what AI can do and so we do a simple demonstration to show how inSTREAM has learned to understand the meaning of content. Using a smartphone, we photograph some articles from a newspaper and email them to inSTREAM and it can tell us what they mean. It can do this because it’s already read thousands of articles like them before on the BBC News website. News articles fit into categories like politics, health, sport, entertainment or crime and by reading thousands of these, inSTREAM has learned to recognise the pattern of each article and categorise them. And it can do hundreds of these every minute.

“But when it sees something it is not 100% confident of or it hasn’t seen before – for example, a cooking story – it will then present it to someone on a screen and say, ‘I don’t know where this fits’. Someone will tell it, ‘This is a cookery story,’ and inSTREAM begins to learn that there is a new pattern and it has a new category name. The same technology can be used to identify different correspondence such as complaints and then to work out how to deal with them.”

Human intervention is still required in the automated process but the dependency on humans is minimised to dealing only with exceptions, and the impact of losing experienced people, or sickness or inconsistent performance is reduced.

The inSTREAM system is already in use by customers including Carphone Warehouse, Kuoni and Davies Group. For travel firm Kuoni, inSTREAM receives over 15,000 tour documents per day from 185 countries and in 40 different languages. Processing all of this information used to require 60 people. It now requires six, despite the fact that volumes have increased by 50% in the last few years. 

Mark Tabone, Director, Products and Services for Equiniti, says: “Business Process Outsourcing (BPO) is becoming more industry specific. What now differentiates service providers is the ability to articulate big ideas. To take financial services as an example, the majority of the large UK clearing banks are wrestling with their costing ratios. Aligned with that you’ve got the new entrants to the market, which are concentrating on products and marketing but don’t necessarily want to carry the fixed cost assets or middle and back office. Clearing banks want to cut their cost base and that’s where we see innovation like inSTREAM playing a massive part. The service providers that can make big dents in banks’ improvement programmes will be the ones that get round the table and work with them.

“Identifying and triaging documents to the correct work queue, as they arrive, is critical to efficiently starting the inbound workflow process and will allow Equiniti to differentiate itself by way of speed and service level agreement (SLA) achievement.” 

The Royal Mail Initial Public Offering (IPO), handled last year by Equiniti, was indicative of the changing nature of communication as the first significant retail IPO in the digital age. Much of it was handled through portals.

“At one point we were taking 6,000 applications a minute via portals,” says Mark. “That doesn’t necessarily change the back office stuff that still needs to be reconciled. It’s just data in a different form and I see inSTREAM being very useful for that. We touch 30 million of the UK population every year and also receive 250,000 notifications of death each year with people needing to be matched with unclaimed assets, so we receive lots of paper and variable media correspondence. So, again, automation and intelligent technology could be a big winner for our customers.”

For more information on inSTREAM visit the Celaton website or contact:

Mark Tabone
Tel: (0)207 469 1915