As part of a discussion series on how innovative companies are dodging data complexity through the use of Synthetic APIs, learn here how, from across many different industries and regions of the globe, inventive companies are able to get the best information delivered to those who can act on it with speed and at massive scale.
Here to explain how improved information integration and delivery can be made into business success, we're joined by Pedro Saraiva, product manager for Content Shared Platforms and Rapid Sourcing at Thomson Reuters. The discussion is moderated by Dana Gardner, principal analyst at Interarbor Solutions.
Below are some excerpts.
You first launched Thomson Reuters content-sharing platform over four years ago after joining the company in 1996. And the platform there now enables agile delivery of automated content-acquisition solutions across a range of content areas. What are you delivering and to whom?
It's actually very simple. We're a business that requires a lot of information, a lot of data, because our business is information — intelligence information — and we need to do that in a cost-efficient manner. Part of that requires us to have the best technology. When we started four years ago, one of the most obvious patterns that we found was that we had a lot of fragmentation of our content acquisition processes: Where they were based, who was doing them, and, more importantly, what processes they were following or not following.
The opportunity that we immediately saw was to consolidate it all, not just around the central capability, but into an optimal capability, with real experts around it making it work and effectively creating a platform as a service (PaaS) for our internal experts in each content area to perform their tasks just as usual, but faster, better, more reliably, and more consistently.
Fundamentally, we are a platform for web-content acquisition. And that is part of our content-shared platform because it's all part of a bigger picture, where we take content from so many sources and many different kinds of sources, and not just web.
I don't know the exact percentage, but I would guess that about half of what we do is content management, rather than site technology, per se. And a lot of those content management tasks are highly specialized, because that's the only way we're going to add value. We're going to understand the content, where it comes from, what it means, and we are going to present it and structure it in the best possible way for our customers.
So, the needs of our internal groups and internal content teams are huge, very demanding, and very specialized. But they all have certain things in common. We found many of them were using Excel macros or some other technologies to perform their activities.
We tried to capture what was common, in spite of all that diversity, to leverage the best possible value from the technology that we have. But also, from our know-how, expertise, and best practices around how to source content, how to be compliant with the required rules, and producing consistent, high-quality data that we could trust, we could claim to our customers that they could trust our content because we know exactly what happened to it from beginning to the end.
Thomson Reuters is a large company. Tell us how large, and tell us some numbers around the number of different units within the company that you are providing this data to.
We have about 50,000 employees worldwide in the majority of countries. For example, our news operations have reporters on the ground throughout the world.
We have all languages represented, both internally and in terms of our customers, and the content that we provide to our customers. We're a truly diverse organization.
We have a huge number of individual groups organized around the types of customers that we serve. Are they global? Are they regional? Are they local? Are they large organizations? Are they small organizations? Are they hedge funds? Are they fund managers? Are they investment banks? Are they analysts? We have a variety of customers that we serve within each of our customer organizations around the world.
And that degree of speciality that I mentioned earlier, at some point, has to take shape. It takes shape in the vast number of different teams we have specializing in one kind of content. It may be, perhaps, just a language, French or Chinese. It may be fundamentals, versus real-time data. We have to have the expertise and the centers of excellence for each of those areas, so that we really understand the content.
You had massive redundancy in how people would go about this task of getting information from the web. It probably was costly. When you decided that you wanted to create a platform and have a centralized approach to doing this, what were the decisions that you made around technology? What were some of the hurdles that you had to overcome?
We were looking for a platform that we would be able to support and manage in a cost-effective manner. We were looking for something that we could trust and rely on. We were looking for something that our users could make sense of and actually be productive with. So, that was relatively simple.
The biggest challenge, in my opinion, from the start, was the fact that it's very hard to take a big organization with an inherently fragmented set of operating units and try to change it, because trying to introduce a single, central capability — it sounds great on paper, but when you start trying to persuade your users that there's value to them in in migrating their current processes, they'll be concerned that the change is not in their interest.
And there is a degree of psychology at work in trying to not only work with that reluctance that all businesses have to face, but also to influence it positively and try to demonstrate that value to our end users was far in excess to the threat that they perceived.
I can think of examples that are truly amazing, in my opinion. One is about the agility that we've gained through the introduction of technology such as this one, and not just the user of that technology, but the optimal use of it. Some time ago, before RSA was used in some departments, we had important customers who had an urgent, desperate need for a piece of information that we happened not to have, for whatever reason. It happens all the time.
We tried to politely explain that it might take us a while, because it would have to go through a development team that traditionally build C++ components. They were a small team and they were very busy. They had other priorities. Ultimately, that little request, for us, was a small part of everything we were trying to do. For that customer, it was the most important thing.
The conversation to explain why it was going to take so long why we were not giving them the importance that they deserved was a difficult conversation to have. We wanted to be better than that. Today, you can build a robot quickly. You can do it and plug it into the architecture that we have so that the customer can very quickly see it appearing almost real time in their product. That's an amazing change.
What was the story behind your adoption of this?
We spent some time looking at the technologies available. We spoke with a number of other customers and other people we knew. We did our own research, including a little bit of the shotgun kind of research that you tend to do on the internet, trying to find what's available. Very quickly, we had a short list of five technologies or so.
All of them promised to be great, but ultimately, they had to pass the acid test, which was evaluation in terms of our technical operations experts. Is this something that we are able to run? And also in terms of the capabilities we were expecting. They were quite demanding, because we had a variety of users that we needed to cater to.
But ultimately, most importantly, we needed the confidence that we could get our job done. If we are going to invest in a given technology, we want to know that it can be used to solve a given kind of problem without too much fuss, complexity, or delay, because if that doesn't happen, you have a problem. You have only partially achieved the promise, and you will forever be chasing alternatives to fill that gap.
Kapow absolutely gives us that kind of confidence. Our developers, who at first had a little bit of skepticism about the ability of a tool to be so amazing, tried it. After the first robot, typically, their reaction was: "Wow." They love it, because they know they can do their job. And that's what we all want. We want to be able to do our jobs. Our customers want to use our products to do their jobs. We're all in the same kind of game. We just need to be very, very good at what we do. Kapow gave us that.
With Kapow, it was a straightforward process. We just click, follow the process that really mirrors a complex workflow in the flow chart that we designed, and the job is done.
In terms of the rapid development of the solutions, it was at least a reduction from several months to weeks. And this is typical. You have cases where it's much faster. You have cases where it's slower, because there are complex, high-risk automation processes that we need to take some time to test. But the development process is shortened dramatically.
We were recently at the Kapow User Summit. We've been hearing about newer versions, the Kapow platform 9.2. Is there anything in particular that you've heard here so far that has piqued your interest? Something you might be able to apply to some of these problems right away?
A lot of what we've been doing and focusing on over the last four years was around a pattern whereby we have data flowing into the company, being processed and transformed. We're adding our value, and it's flowing out to our customers. There is, however, another type of web sourcing and acquisition that we're now beginning to work with which is more interactive. It's more about the unpredictable, unplanned need for information on demand.
There, interestingly, we have the problem of integrating the button that produces that fetch for data into the end-user workflows. That was something that was not possible with previous versions of Kapow or not straightforward. We would have to build our own interfaces, our own queues, and our own API to interface with the robo-server.
Now, with Kapplets, it all looks very, very straightforward, because we can easily see that we could have an arbitrary optimized workflow solution or tool for some of our users that happens to embed a Kapplet that allows a user to perform research on demand, perhaps on the customer, perhaps on a company for the kind of data that we wouldn't traditionally be acquiring data on a constant fixed basis.
Any advice that you might offer to others who are grappling with similar issues around multiple data sources, not being able to use APIs, needing a synthetic API approach?
I suppose the most important message I would want to share is about confidence in technology. When I started this, I had worked for years in technology, many of those years in web technology, some complex web technology. And yet, when I started thinking about web content acquisition, I didn't really think it could be done very well.
I thought this is going to be a challenge, which is partly the reason why I was interested in it. And I've been amazed at what is possible with technologies such as Kapow. So, my message would be don't worry that technology such as Kapow will not be able to do the job for you. Don't fear that you will be better off using your own bespoke C++ based solution. Go for it, because it really works. Go for it and make the most of it, because you will need it with so much data, especially on the internet. You have to have that.