In our first interview with Arria CTO, Robert Dale we learned how powerful a tool NLG is when looking to automate the task of turning vast quantities of data into meaningful text. The range of applications for business is huge, and Arria is already working across a range of verticals with customers spanning Oil & Gas, Advertising, Utilities, and Financial services to help businesses make sense of their data in real-time.
The most exciting part of the story however is yet to be fully realised and given the new developments Arria is working on, we were excited to catch up with Co-Founder and CSO, Matt Gould to find out what lies ahead and the ultimate goal for the NLG roadmap beyond report writing.
Matt, we’ve given a brief overview of Arria NLG for our readers who might not have seen the previous interview, but from a Co-Founder’s perspective could you elaborate more on the big picture mission for what you’re trying to achieve?
Arria NLG’s vision is to be the global leader in the commercialization of Natural Language Generation (NLG) technology. We’re producing an artificial intelligence that didn’t exist before, as it has one of the most fundamental human skills – the ability to communicate. Today, corporations are constantly competing in new spaces, and are often unable to continue their growth trajectories on the back of human capital alone – that’s where our NLG platforms are providing much-needed solutions. Infinite amounts of data are being generated every single day, and the lack of experts with the ability to analyze this data has become apparent. Our mission is to continue using NLG and automation to innovate business processes and free the best and brightest to do what they’re trained to do – engineers to build, doctors to heal, scientists to discover.
What is it that really differentiates Arria from other NLP specialists we come across?
Arria NLG’s platform, unlike others in the field, is not text-to-text. Our platform mimics an unconscious process in the human mind and replicates it. Our platforms are tailored to analyze data as an industry expert (e.g., doctor or weatherman). We put the wisdom of the expert into the machine and it understands it just as the expert would and communicates an insightful report in natural language. Essentially, the Arria NLG platform gives a voice to data.
Other companies in the field are using a template system akin to mail merge on steroids. The template system requires that everything has been written before and someone has already envisioned the outcome or response. Arria’s platform, on the other hand, can explain data it has never seen before. It can also identify what information is important and then tailor the language based on audience. For example, you might be visiting a loved one at the hospital and all of his or her medical information is in a medical chart at the foot of the bed. You don’t understand the technical terms or abbreviations because you’re not a doctor. NLG platforms could review the medical chart and generate a report for you (a non-medical professional) in laymen’s terms to explain your loved one’s health, from diagnosis to treatment.
With offices spanning the globe; from London to Auckland – where do you see the tech innovation coming from with respect to AI?
Arria’s global band of scientists and engineers are building AI technology that is helping to herald the AI-human hybrid intelligence revolution, where people and AI systems work together symbiotically. The future of tech innovation is about finding a tighter integration between humans and machines, not a replacement. That’s why we’re increasingly building systems that support human and machine co-authoring of texts. The machine writes the bits it can do well, and we make it easy for humans to augment this, adding further insights or revising what the machine has written. Alternatively, the human author can be the primary author, calling on our modular NLG services to provide fragments of text if the author wants to delegate those writing tasks. The result: NLG co-authoring gives you the best of both worlds. Human authors bring their insights and nuance, and their subtle understanding of audience. Machines can do the grunt work that would otherwise take a human author endless amounts of time, if it’s feasible at all, delivering detailed and accurate descriptive narratives about the information that would otherwise be left buried in data.
We’ve seen the platform in use in many vertical applications – since we last caught up with Arria, where do you see the biggest potential?
The biggest potential is in adding narrative layers to many of the leading application and cloud services currently available. True two-way communication via language will be possible between applications and people. Here are some examples:
|USE CASE||IMPACT||DATA POINT|
|E-Retail: Currently almost all product descriptions on the internet are written by people; they are static – the product description is the same for any shopper on the internet. Arria’s NLG Developer Cloud may allow e-retailers to personalize product descriptions for each customer in real time (millions every second), combining the data in retailers’ product database with what online analytics tools such as Google know about each customer to provide a personalized service on every product page.||For the first time customers will be presented with products and services in a way that prioritizes the individual’s interests, demographic, and psychographic. Personalized product descriptions drive a higher conversion rate.||Amazon sold 488 million products in the USA alone in 2015. Almost all of those products were written up by a person. The Arria NLG Developer Cloud could write the bulk of these automatically, and in real time. Product feature changes? Adjust the underlying database once and the NLG Developer Cloud will update the product description for every site it is offered on.|
The NLG Cloud allows developers to use more than templates or pre-recorded phrases if they want their applications to write reports or speak naturally to users. By using the NLG Developer Cloud developers can extend a narrative layer to any application they are building. Health apps can be given a virtual doctor, sports apps can be given a virtual commentator, stock apps can provide a personalized written report in seconds, etc.
|For the first time, applications and the Internet of Things will not only be good at hearing and understanding the natural language of people (voice capture) but also writing and speaking back to us using language that has been built from scratch in real time – a natural, swift process for the human mind, until now almost impossible for an artificial one.||There are approximately 18.2 million software developers worldwide; there are approximately 1.5 million applications in the Apple App Store; there are approximately 1.6 million applications in the Android store. None of these applications currently possess the narrative and natural language generation technologies that Arria has patented and will be providing via the NLG-Developer Cloud.|
There had been a tendency among some of your early customers to opt for ‘on-premise’ deployments with concerns over privacy of data – have you found a mind-set change toward the cloud based services more recently?
Yes, many of our customers are finding it more convenient to use our cloud-based NLG service. Genpact’s Systems of Engagement portfolio, for example, embeds leading digital technologies within cloud-based offerings added on top of clients’ existing systems to streamline and transform operations in an agile and flexible manner. The NLG Platform via the cloud allows for swift, simple capture of expertise, and then running a narrative reporting or communication layer back into the client’s existing enterprise environment. There are no long-winded integration exercises or problematic intersystem configuration exercises.
Looking beyond the capabilities of advanced report writing/decision support – what developments are you working on for Arria’s technology?
We’re working on true artificial intelligence language – some of the seemingly simplest effects of our technology required the deepest levels of coding and effort; to take one example, Quant is a tool we have built that describes any Excel chart or graph in language – you literally highlight the associated value cells or the Excel chart itself and Quant will write 50-100 words to describe it. Now imagine Quant as the voice of an artificial intelligence describing a phenomena it is seeing in any data. Think Jarvis, the AI from Iron Man calmly telling Tony Stark about anything – from his heart rate to the trajectory of a missile in flight. Quant makes that possible in real life, not just in movies.
How would this fit into the Enterprise setting?
In the Enterprise setting there are two effects. First, precious expertise that may be causing a log-jam in critical processes (think experienced electrical engineer at a power utility) can be enrolled into the NLG Platform – the platform starts looking at the enterprise systems data as the expert does, and reporting on phenomena as they do – but faster, and at volume, and never sleeping or eating or retiring. Big productivity gains.
Secondly, the NLG Platform can speed and automate critical insights that were previously too time-consuming to uncover. Take any online retailer trying to manage all of the channels and marketing points of contact their brands are interacting with – the sheer volume of Google Analytics, CRM analytics, and payment analytics is just too large to deliver insight and recommended actions in real time. The NLG Platform can do that heavy lifting for the brand manager – distilling all of the data down into a set of simple, narrative recommendations (e.g., “Do this to improve sales by 5% or do that to cut costs by 8%”) that anyone, even a non-expert, can follow and use to take action.
Is this changing your target market or simply broadening your scope?
Broadening our scope. Our latest patent applications and technology developments are being applied to our growing list of global clients, as well as to a range of SaaS-delivered packaged applications that are in development for a number of markets. The patents also apply to our proprietary Software Development Toolkit. Arria’s expanding patent portfolio strengthens our plan to become the global leader in automating data analysis and information delivery in any industrial or consumer setting.
What market barriers do you need to overcome to get the early buy-in?
The biggest challenges to get early buy-in are related to change management. Large enterprises often flinch at the thought of the change management required for introducing new software systems. The processes that companies use have been tried and tested over many years – introducing new systems, no matter how useful they may be, takes time and care. Fortunately, the NLG Platform embodies an organization’s existing culture in its artificial intelligence, so it’s a solution swift to implement culturally. Our clients tend to revel in the practicality of the NLG platform – nervous at first when they see it writing and reporting as they currently do, but effortlessly, swiftly, and with immediacy they tend to smile.
If you had one message for CIO/CTOs of large scale Enterprise when looking at the opportunity AI holds, what would this be?
Cars are not the only things being made smarter by AI systems and cognitive computing. The AI revolution can help large scale enterprises derive greater margin from their operations, improve safety, and customer satisfaction.
Here are some examples of systems in place already:
|One of the largest Brokerage Firms in the USA||This globally recognized brand uses the NLG Platform to produce written, personalized portfolio reports for its online clients. Before using the NLG Platform, the client was restricted to sharing graphical updates, or quarterly, general reports written by a person – NLG allows better service, personalized service, and swifter service.|
|One of the World’s largest Travel Distribution Systems||The world’s favorite travel marketplace wanted a system that allowed travel itinerary choices to be provided over the phone in natural language. They chose the NLG Platform. For users who want to talk to virtual travel agents, and not just click menus on a screen, there is now a better option.|
|One of the World’s top three Accounting firms||This client was looking for a way to reduce the time and human resource cost of generating quarterly compliance reports for its thousands of clients in Europe. Each report could take up to an hour to produce, and was a burden to the highly-trained accountants who had to draft each one. Now, thanks to the NLG Platform, a client’s data can simply be ‘click-dragged’ into an online portal, and the full audit report is generated in seconds.|
|One of the world’s top three Fast Moving Consumer Goods suppliers||This corporate giant needed a way to produce swift, insightful analysis across a huge range of products, divisions, and geographies instantly. The NLG Platform was configured to embody the knowledge of the company’s financial analysts. The multiple-page reports, replete with tables and charts, could now be requested in real time – and they look and read just like those it used to take days and weeks to complete.|
|Europe’s largest private Weather Service||Reducing the work of hours to seconds, the Road Ice Reporter for this client produces written weather reports in both English and German. The NLG Platform can be supplemented to provide output in a variety of languages – same data, different audiences.|
|The world’s largest technology services company – Oil and Gas||This partnership-led product built using the NLG Platform provides written updates in support of leak detection at refinery and processing sites. Called POLUS, and incorporating IBM’s Watson technology in addition to Arria’s NLG Platform, it acts as a virtual site engineer, tracking issues and matching them to emissions-control legislation for remediation.|
|One of the top 25 global brands by value in the world – Financial Services||This household credit card provider needed a faster way to generate financial model validation documentation – a critical compliance requirement for such organizations. A heavily human-dependent process requiring document drafting and multiple authors and inputs could be largely automated thanks to the Arria NLG Platform. A core automation saving at the heart of transaction businesses.|
|One of the top two Cable Services in the USA.||This client wanted to provide real-time personalized usage statements to its millions of customers across the USA – customers who rely on their cable provider for not just entertainment content, but their broadband and telephone services in combined bundles. The client chose the Arria NLG Platform to do the heavy lifting within this critical customer care service.|
|One of the largest Investment Banks in the Asia Pacific region||Private wealth clients are a large part of this client’s business. Providing timely updates on the state of each customer’s portfolio was an increasing burden on the expert analysts at the bank. By incorporating aspects of those experts’ intelligence into the NLG Platform, the bank was able to provide real time updates to its thousands of clients globally.|
Full details can be found here on the event and who’s already signed up can be found here – http://theaisummit.com/london/attending/ If you’d like to find out more on how you can leverage NLG and the broader opportunities AI truly holds for the enterprise then you could explore this and much more at upcoming The AI Summit, taking place on the 5th May at the Four Seasons, Park Lane, London. Arria is presenting an exciting case study which dives into the real world impact of the technology from a business impact perspective.