Dave Tyler: Nuxeo was one of the first companies to rally behind Gartner's Content Services approach: how has the market responded to that message?
David Jones: In recent times we've had almost a two-tiered go-to-market approach, where 2018 has been focused for us on 'interpreting the Content Services mantra'. Since Gartner came out with that concept we've been firstly figuring out how our technologies apply, and then creating a message based on that understanding. At the same time we've been working on developing our own specific technical excellence within the platform itself, to ensure that we are in the best place possible moving forward. I genuinely feel we are ahead of the entire market right now.

We've spent a lot of time looking at Machine Learning and AI, as well as federated connectivity. A lot of people say that they federate content, but what they mean is that they allow users to pull a document from Box or Dropbox - or Alfresco or Documentum - into the system. Our approach is that you don't even need to 'pull' that content from anywhere; it's more about knowing where it is and how to access it, how to get metadata from it.

DT: What exactly does that mean, and how does it work in practice?
DJ: Crucially, we're talking about a two way link: if we change the metadata within Nuxeo, that can be pushed back to the Alfresco/Documentum repository. Equally if that gets changed within Documentum because they're using that (and applications built on top of that), Nuxeo is automatically notified of those changes. That is true bi-directional federation, and very few people are able to offer that. And that is key, otherwise you just end up with systems that are duplicating your data - again!

To manage that you have to go back to the old style of ECM: the 'one repository to rule them all' approach. Which doesn't actually take anyone any further forward; it leaves you with one dirty great big repository, instead of an intelligent one that stores the metadata and index information, while leaving the actual data where it should be - at least until you're ready to move it.

DT: It seems that your focus is as much on the end users and their data requirements as it is on the system owners/administrators; is that correct?
DJ: It is what I have described more than once as 'an outbound way of looking at content management': it used to be that you had an interface via your ECM system and that was how you accessed your documents. That was how we had to do our workflows, and frankly it was a nightmare. The interfaces were designed for DM/information management professionals, and they were horrible! They were of no practical use to the average Joe who just needed to access a document. As a result those solutions were never really used outside of that core team.

We're looking at how to provide a service to the business: if Salesforce wants access to documents, or Outlook, or they need to build a mobile app, we give people one single easy access way to get that information - wherever that data happens to be held within the organisation.

DT: So Nuxeo serves almost as a middleware layer between user applications and any number of disparate repository-based systems - completely transparent to the user?
DJ: It's a bit like a hub and spoke approach, as Nuxeo acts as the central conductor or processing piece for all those systems - but also adding value. For example, it can be very difficult to change metadata schemas in many legacy systems: if you decide you need to store some additional data against a customer, you might have to almost redesign the whole thing from the ground up. It takes time and effort, and can be a considerable pain. We work on the basis that you can do that much more simply within Nuxeo, and expand the schema in any way that you want.

This is something we're doing a lot with customers; connecting into legacy systems that may have been in place 20 or 25 years, running them through AI-based classification. There wouldn't have been any OCR initially, so no-one has any real idea what is in these documents. We can connect to these systems, pull documents out, run them through this AI-based OCR or classification tool, so that when we put them back, we have this added layer of additional metadata that can give users so much more information about what's actually in there. That can be a huge help to a business in deciding what you need to keep, what to get rid of, what needs securing, and suchlike - which is key with GDPR now being in force.

DT: You've touched on AI and machine learning - are these technologies developed in-house or are you OEMing existing tools?
DJ: When we started the whole process we were connecting to Amazon and Google AI offerings - which we still use where appropriate - but in addition our own R&D team in Lisbon is building modelling capabilities based on those technologies to add functionality. Looking at the automotive industry as an example: knowing that a specific document has a blue truck in it, is of absolutely no value to a user. But training an AI-based recognition tool with domain knowledge (such as that manufacturer's product catalogue, for instance), will allow the system to recognise a particular model, variant, engine size, optional extras - all of this gives the user a vast amount of additional context.

The framework that we are developing will take those 'raw' algorithms, combine them with domain knowledge, and build training mechanisms for business-specific models. That will be based on our own IP and we're very excited about it.

DT: What is the strategy going forward for the rest of 2018 and beyond?
DJ: The focus for us right now is all about continuing to develop that technical excellence to build on our strengths. Our Connector framework lets us plug into any new technology; that's how we connect to the AI tools, for instance, but also to capture tools, Records Management tools - whatever the business requires.

Specifically for 2018 the focus has been on how we got to market, and how we communicate our message to the marketplace. When people say Content Services Platform (CSP) is a replacement for ECM, the natural response might be "Well, so what?". We have to make the story relevant to the real world. We're talking about modernising - without adding complexity - and it's very definitely not about ripping and replacing. We know that organisations have so many different disparate systems, not connected together. The problem isn't ECM: the problem is that organisations simply cannot tie those systems together as they need to, in order to use their data efficiently as the basis for their information management. That's where Nuxeo comes in.
More info: www.nuxeo.com