 in Boston, Massachusetts. It's theCUBE covering HPE Big Data Conference 2016. Now, here are your hosts, Dave Vellante and Paul Gillins. Welcome back to Boston, everyone. This is theCUBE, the worldwide leader in tech coverage. Darren Harris is here, he's the CEO of HTI Labs. He's joined by John Glass, who's the CTO. CEO and CTO, we got the right people here to talk to, gentlemen, welcome to theCUBE. Thank you, it's great to be here. Yeah, so we were just talking offline about your relationship with HPE, you know, formerly HPE. You've been working with those guys for a while, I guess, but so tell us about this event. You're over from London? Yeah. What do you got going on here? So we're here to demonstrate kind of the extension to our product. We've made our product school schematic and we've extended it to access the HPE Haven on Demand functionality, the APIs that expose the applied machine learning APIs. So schematic our product, it's an Excel-centric product from a user interface, but we extended to handle unstructured data, solve a lot of the technical complexities and scalability issues that you have with Excel at the box. And what we've really done is taken the Haven on Demand APIs one steps further than just simplified for users to access and basically makes that any user of Excel can access those APIs, experiment with them and run unstructured analytics against the Haven on Demand backends. So John, you've built a platform that extends to Excel. So there's a user, I see Excel, but then you've got your back end behind it? Yeah, absolutely. So Excel is really the sort of premier business analysis tool for people that are in all kinds of industries looking at all kinds of data. But people do struggle with data volumes and structured and particularly unstructured and semi-structured data. So our product addresses that. It keeps the core Excel experience that users love, but it allows them to work with bigger data sets. It allows them to offload, processing to things like Vertica and now Haven. So people can suddenly process data that they didn't realize they could still deal with in Excel. So what's that process like then connecting into the Haven APIs? I mean, it's really as simple as it's a function. So we have people that we're working with who have been using a VLOOKUP or a FIND function in Excel to try and get something out of a piece of text. And that's really hard. If you have medical trials data and you're looking for drug names, they're written in a sort of weird kind of Latin. There are lots of variations. There are lots of misspellings. So what we're kind of saying to people is take your spreadsheet where you've got FIND, you just replace that with Haven.extractentities. It's really as simple as that. So we're trying to make the bar to entry as low as possible. The Haven team have done a great job of making these APIs really well encapsulated, really well described, really easy to understand conceptually. And we're trying to remove the technical barrier and give them to end users as well as app developers. So you do the Haven entity extraction, present that in Excel and your eyes don't bleed. Exactly. What does it look like to the user? And what kind of function, to an Excel user, what kind of power are you giving them that they didn't have before? So the core IP, I suppose, of what makes it different is that we have within a cell what we call a data link. And that data link is a pointer to unstructured data that kind of sits outside of the core experience of mapping everything down to rows and columns when you work with it. So when you look at Excel, what you see is a viewer on the right-hand side that allows us to visualize these data links and work with them. So users can perform operations with other functions over the data links and kind of build up a whole dependency graph of calling different functions and then debugging into the viewer and then collapse that data back into Excel once they're finished with it. Tell us more about the company. How long have you guys been around? So, ACI was founded in 2012. The core backbone of the team all worked together at an investment bank in the UK. We've been working with spreadsheets and front-office risk systems for years. We're a member of the London Stock Exchange Elite Program, technology program for fast-track growth companies. And yeah, we're primarily doing a lot of work for energy trading companies. Anybody that really has Excel and they've got into an unwieldy mess or they've got some structure or they're lacking governance, which is a real key thing. We work with them. We leverage schematic as an accelerators platform and we help them simplify their estate. They may have stability problems or complexity problems or lack of transparency problems. And for us, where we interface into pricing APIs or in-house risk systems, accessing a REST API such as Haven on Demand is just a natural step. And really the goal is all about putting the tools directly into the hands of users, removing the bottlenecks from core teams and allowing the users to innovate and understand the data. Well, they understand the data more than just depending on technologies to interpret that. That's kind of the holy grail of big data, isn't it? As opposed to insights for the few, which is sort of the big criticism of decision support and BI. And even big data with data scientists. You got a data scientist if you can find him or her. And so this notion of citizen analyst. So you're seeing that actually in the real world. Yeah, I think there was so much complexity in big data, all the tool sets, all the components, and there's just a sea of things. And what you see with, I suppose, Haven on Demand for the first time is the packaging and the robustness and the simplicity is making it so much more accessible. And interestingly enough, when you start to consume from the analytic, you're agnostic to the underlying storage platform. You have connectors, it goes in, it could be MongoDB, it could be Hadoop, it could be whatever. But they've really lowered the bar. In terms of if you could literally subscribe for an API key, you could access this stuff instantly and straight away you could do sentiment analysis. We could take this broadcast, convert it into text and run sentiment from a sheet and do that, and anybody could do that, but could use Excel. That's, well, we saw the video today, Robert Young-John sort of teased us with that, the ease of use. And you're saying it's really that simple. It's robust, and if you could confirm that, and then I got a follow-up question. Yeah, I mean, it's really straightforward. I think one of the nice things the Haven team have done is Idle is a big, powerful platform. And in the on-prem version, there are many hundreds of APIs that you can use. So actually there's a skill involved in distilling that down to things that are simple and very clear and very easy to use. And that's kind of what we like is it's something that people can self discover. They can work with the individual API functions. There's also combinations, which is launched this week, which we've been part of. We've got support for that now, kind of already in schematic, which we're quite pleased to get out there first with. And that means that people can put these things together. They can chain operations together and do things like language agnostic sentiment processing, which is one of the combinations that's been launched as an example. So it's a simple idea, but you feed some text into the language recognition module. That tells you what language it is. You then take the knowledge of what language it is, and you feed that text into sentiment analysis. So you can take any text in any language, and you can do sentiment analysis on it as a single call from a single cell in your spreadsheet. So you've got this platform, you license this platform as a subscription? Yeah, so there's, I think Mike said, there's four components. There's the workbench, which is on a per user license. And depending on if they want to, you know, manage and govern the assets. So effectively think of it like source control and all the best software engineering practices applied to spreadsheets. We've kind of built a manager. And we're also really big on capturing usage metrics to allow transparency on what people are doing and how they're doing it for kind of core people to govern that, like core IT teams. There's a server where you may offload. You can define effectively calculations or algorithms from Excel and emit those as code or batch files, or batch files, but actual code to run on the server, part of a batch. And the portal is all about sharing it. So you can literally take a sheet, use a haven on demand API and publish that to a web portal and view that on a phone. So you as an Excel user can go right through the whole journey and other people can take that because an input cell in Excel is the same as a text that it box in a form. So you can effectively define a whole calculation tree and then push it out for consumption. Are you able to access external data from within Haven or is the user supplying the data and you're supplying the processing? So we're, we can, you know, we have a number of data sources that are supported kind of out of the box by schematic and we're constantly adding new connectors. So all of the things that you would probably expect, CSV files, XML, JSON, web scraping, web APIs. So, you know, grabbing HTML data, grabbing XML, connecting to relational databases. So, you know, Vertica is one of the relational data sources that we support and essentially it means that people can pull everything together into Excel. They have a view onto a table that's held remotely. It might have, you know, hundreds of millions of rows or more in it. They can have a sort of remote view onto that table and they can work with it and pretend that it's there, you know, on the desk and it's actually, you know, all happening on the server. But when they're ready to start combining things together, you can create these mashups, you know, and the real power and the real value comes from the local data, the local keys and things that you have available and the big, you know, the big data sources that you've got access to remotely. So you licensed this by user? By user and then by module and then the components are kind of related to the, you know, how far deep they go. And then you just sort of bundle in the whatever, haven costs associated with that. Yeah, this is essentially what we've actually done is we can, the APIs that we expose as we call them first cost functions into Excel, they themselves can be defined in sheet templates that are then pushed and compiled and then pushed back to other users. So the idea that a power user of Excel could define a new API and push it to other users to consume. So really looking at the whole way is how can you empower non-programmers to self-publish, generate new APIs and share them and abstract the complexity from other users that might not be so advanced. So your platform will discover Excel on my client and then you can just install it. So the minimum barrier is one installation on one desktop machine and immediately, all the data sources that you have access to, that anything that we support, you can get hold of. We write custom connectors for our clients as well. So lots of clients have legacy systems that are often quite painful to operate with or they have internal APIs that they use. And we can give analysts, risk managers, people doing the line of business work access to these data sources. And that has a really nice effect. It makes them more effective but it also makes the systems better because people start playing with the systems. They start trying, what happens if I do this? What happens if I do that? And you get a really nice feedback loop with the developers as well. People get very excited about the fact that they can push new versions, their users can try them. They can actually iterate the development process faster as well. So are you enabling this through a community or through a code exchange type of platform? So at the moment we've kept it with corporates and there's downloads. We haven't really pushed it. We've been really focusing the last couple of months on adding Haven support and the core infrastructure but our strategy is to get the message out there now really and start to put it in lots of people's hands. HPE said that they have 70 APIs in Haven right now. Are you looking, what are your next extensions you're looking at? So we're looking to extend support to allow combinations to be created in from Excel and then pushed up. Can you give an example of a combination? So that would be the language independent sentiment where you have the 70 APIs and they're quite powerful in their own right but when you start to combine them and you start to have- So recognize the language and then determine sentiment. Yeah, that's correct. So that is an example of a combination. This is a key part of API design I guess is have units of functionality. Don't try and do too much in one place. So as IT developers, internal dev teams should be providing units of useful functionality to their users and leaving the users to creatively combine them. And I think when applications that people use start to presuppose the way that those units of functionality would be combined you actually lose power, you lose capability. So that's part of our philosophy and it's one of the things that's really drawn us to Haven because we think that the units of functionality that Haven on demand provides and this combination framework together is just hugely powerful, it's really exciting. And you license Haven by the function that you- So Haven is licensed on a usage structure? For the whole- For the whole, yeah, for based on how frequency. So what's great about it is there's a free tier that you can actually go and get quite a lot of functionality out of the moment. So you could download out trial schematic, you could download, you could get an API key for Haven and you can just start today, you could, you know. And just to give you an example of some of the power there was a report in the UK recently about the Iraq War with the Chilcot report. It's 2.2 million words in size, three times the entire work of Shakespeare. With that, with Excel, we were able to analyze that, look at notable people that were mentioned, rank them by the number of mentions and sort that with a reference of where they were mentioned through the document. And that's just one thing with one document that somebody could do at home with their PC now. And I don't know of anything else that, you know, a combination of that power that people can do that. Are you at the scale where you're having to pay for Haven yet or? Not at the moment. We're not sure, we think they might be being kind to us. Yeah. Yeah, we don't want to mention that in case somebody's going to give us a big bill at the end. I want to see if I can get that deal. We have a need for this. Now why Haven? Why not? Did you look at Watson APIs or other platforms? So I think that, well, why a big company, I suppose, and then we talk about the other bit. So there's a lot of stuff out there that might be open source or libraries and things like that. When you deal with enterprise customers, you want to back onto something, you know somebody's going to be behind there in a robust way with the enterprise. So there's a lot of really cool stuff that's in research, but with our enterprise type customers, we kind of have to have that. In terms of knowing a bit of background about the idle platform, there's a lot of power under the hood. And really, from our perspective, we believe that this is only just scratching the surface of the capability that's going to come. So Watson, we can look out, but I don't really think they've kind of got us advanced yet in terms of the APIs that they're exposing. I'm sure it's a gap they're looking to close, but like I said, knowing what's under the hood in the autonomy engine or the idle platform, we think there'll be a lot of innovation coming out. Had you worked with idle prior to the HP acquiring? So I had colleagues back in when I was in the industry and I knew very much about what was going on with analysis of unstructured data, especially around trade confirmations, legal documents and the use cases. So we were a lot more in the structure pricing, risk management, Monte Carlo farms and Cal Grid and Space. What about visualization? How do you handle that? Is it through Excel or do you have another visitor? Yeah, our core product is schematic, it runs in Excel and Excel has some very good visualization capability. So really anything that you can do in terms of visualization in Excel, you can take data that you've run through schematic functions, you can pull that back out into a sheet, you can do charts, you can do graphs. Part of the portal aspect of our platform is an ability to surface the same content you'd see in Excel onto any device across the web, but also the ability to create, effectively create new APIs. So if I create a report which is a data set with some inputs that drive what that output is, I can surface it as what would look like a spreadsheet or some charts, but I can also surface it as an API which can then be ingested into another tool. So for the really advanced dashboarding solutions that's kind of where we're going is to sort of pipeline through into other systems. And how many are you today? Is 12 for us? Where do you want to take this, give us the vision? So, well, we see such efficiency games. It would be something that anybody with Excel that has challenges and hits problems and stability issues would start to use really. And I think where we'd like to take it is add more connectors, give put them in the hands of people and allow them to really drive innovation. There's lots of talk about big data and big data technologies and how we solve that problem, but there's still a lot in that kind of small and medium data space where complexity hasn't really been solved. Engines are still not being serviced and there are massive projects that are kind of not delivering and not delivering real value. There's a lot of problems today that people have with working with data and we really like to help people tackle those and kind of control that and this complexity is still there. Well, Darren and John, thanks very much for coming to theCUBE. We'll see you probably in London in December. We'll check out the DeLorean and the hoverboard. What should people Google? It's the Google London hoverboard. It's the first video. Check it out. It's quite amazing. These gentlemen were behind it. All right, thanks very much for coming to theCUBE. Thank you. All right, keep right there, everybody. We'll be back. You're watching theCUBE, the worldwide leader in live tech coverage. We're live in Boston. Be right back.