OpenText looks to navigate crowded machine-learning analysis waters with Magellan

OpenText is entering the sizzling but crowded machine-learning analytics sector with Magellan, an Apache Spark-based offering that draws on the company's $330m acquisition of enterprise reporting and analytics stalwart Actuate in late 2014, as well as its purchase of text analytics vendor Nstein Technologies a few years earlier. Magellan is a platform aimed initially at data scientists and developers but ultimately at business users as it gets integrated into OpenText's existing business application portfolio to enable machine-assisted decision-making, automation and business optimization, as well as being sold as a stand-alone offering for advanced analytics.

The 451 Take

Announced a year ago and delivered on time, Magellan represents a major investment for OpenText – both organic and inorganic – and sees the company taking advantage of the deep understanding of unstructured data it has built up from its roots in search and content management. Given its focus on content management, which is different from partners such as SAP, Salesforce and Oracle, it should be able to avoid them for quite some time in competitive situations. IBM Watson is a different matter, perhaps, but we don't suspect that many customers will find themselves choosing between the two. Magellan should also expand OpenText's developer network, which is something it needs to do, as machine-learning functionality will attract developers. It is investing in a new developer education program to help make that happen. There are improvements to be made in terms of usability and data management capabilities required for advanced analytics. And as every major software provider now has a machine-learning strategy of sorts in place, Magellan might one day not be seen as quite the differentiator it may currently appear to be. But right now, Magellan represents a logical starting place for OpenText's customers to infuse their applications with machine-learning smarts.

Context

OpenText is perhaps best known for its content management systems. But the company started down the path of adding analytics – and therefore intelligence – to its product portfolio in 2010 when it nabbed text analytics vendor Nstein for $33.5m. OpenText, which continues to use M&A as its leading growth driver, subsequently snagged Actuate in part to add analytics to its portfolio of business applications. The company is now looking to add a new level of intelligence to its business apps with Magellan, a machine-learning, BI and advanced analytics platform that it introduced at its Enterprise World conference last year – billing it as the next phase for the company – and has now unleashed.

Strategy

Magellan is designed to be the cornerstone of OpenText's transition from a supplier of mainly enterprise content management (ECM), business process management (BPM) and customer experience management (CEM) applications into a purveyor of a more intelligent platform and apps. In other words, the company is using Magellan to instill machine-driven intelligence and analytics into its entire portfolio and in so doing underpin its goal to place advanced analytics into the hands of every user.

Existing OpenText users employing its ECM, CEM, BPM and Business Network apps will be the main focus for the time being. Deploying Magellan as a weapon to win new business in analytics is the company's other go-to-market strategy. It will be pursuing organizations with at least $250m in revenue and 500 employees. OpenText is therefore aiming to upsell into existing accounts – including those already using its Actuate-based analysis capabilities – as well as employ Magellan to win business in the advanced analytics sector.

The company is seeking to use Magellan's delivery model as a preinstalled appliance on standard x86 hardware on-premises, in the cloud or as a virtual machine as a key differentiator from other offerings of this ilk, such as IBM Watson and SAP Leonardo. It believes that Magellan can be deployed on a standard hardware stack costing no more than $100,000.

OpenText is also looking to single Magellan out from the crowd through its open foundation, which is designed to make it extensible and flexible. Magellan uses Apache Spark's MLlib library of machine algorithms and also draws on Spark for data processing. Additionally, Magellan employs another open source offering in the shape of the Hadoop open source data-processing framework. Hadoop is a prerequisite in the initial release, as it is where all of the content Magellan analyzes and learns from resides.

Integration with Jupyter Notebooks is another example of the company's strategy to make Magellan open. Jupyter Notebooks is employed by data scientists and developers for advanced analytic use cases as it supports some 40 programming languages, including R, Python, Scala and Julia. OpenText is also tapping into the offering to address the all-important area of data preparation. Any other data management issues that an organization might have, such as data quality, will be addressed by the company's professional services team, which will also help with other issues like model review.

Products

Magellan unites OpenText's acquired Nstein functions for text analytics and auto-classification with capabilities from Actuate and its Quiterian pickup for visual predictive analytics. Actuate reached for Quiterian in 2012. Other core components – aside from Spark for data processing and machine learning – are connectors to hook into data sources requiring analysis, as well as an SDK framework to build others. Magellan will have some 30 connectors from the get-go, with many more to come.

Data scientists kick off the machine-learning analysis process by hooking into Spark MLlib using Magellan's Machine Learning designer tool – or by crafting their own machine-learning models in Magellan Notebook (based on the open source Jupyter Notebooks). OpenText Big Data Analytics and Information Hub (iHub) – which are integrated – are already pitched as high-performance big-data and predictive offerings for business analysts and are other core components. Users can flip over to this product and apply machine-learning models to the data and perform model execution as a one-time task, or scheduled on a recurring basis. OpenText Big Data Analytics also houses statistical techniques for data profiling, mapping, clustering and forecasting, as well as features for creating decision trees and association rules, and doing regressions and correlations.

Additionally, OpenText Big Data Analytics brings data discovery and visualization to Magellan. The iHub component is deployed to publish a visualization as a data object for use in a dashboard that can be viewed by a business user, as iHub is an environment for designing, managing and deploying dashboards, reports and analytics.

During the launch at Enterprise World, OpenText demonstrated four scenarios employing Magellan. The first was target marketing for campaign management, the second was forecasting based on weather patterns using linear regression, and the third was predictive fleet maintenance deploying a random forest algorithm. Those three all involved data being taken out of the application to have intelligence added to it by various machine-learning algorithms. But the fourth was OpenText People Center, an HR application built on OpenText AppWorks, its low-code development environment that had Magellan (in this case, logistic regression algorithms) built into it. That model is the eventual goal for all of the company's applications.

OpenText plans to deliver updates to Magellan every six months after launch. Magellan 1.0 was unveiled on July 11 and made available as an appliance. Availability on the OpenText Cloud as well as third-party clouds is in the works. The company is currently working on serving it up within its datacenters in the US, Canada and EMEA. The intention is to ultimately make Magellan available in all OpenText Clouds, third-party clouds, and on-premises.

Providing tooling to make Magellan more user-friendly has been earmarked as a roadmap item. This will include simplified administration and management. Supporting industry-specific use cases is also in the cards. The company plans to concentrate on the healthcare, finance and banking, government and automotive verticals. OpenText's broader vision is to integrate Magellan across all of its product lines.

Competition

Although OpenText's initial focus is on its installed base, which makes it less susceptible to competitive threats, the company made it clear that it intends to take on IBM Watson and other machine-learning environments peddled by big guns – such as SAP Leonardo – with Magellan. We also think Magellan is reminiscent of Oracle's Adaptive Intelligence Apps, Salesforce Einstein and SAS Institute's strategy for its Viya cloud-friendly platform.

OpenText's positioning as a stand-alone advanced analytics environment built on Spark is likely to elicit comparisons with H20.ai, which also peddles openness courtesy of its open source underpinnings. Dataiku is another machine-learning player with a strong focus on open source tools, which it employs as the foundation for advanced analytics. Additionally, organizations seeking an open-source-based machine-learning environment could consider RapidMiner, KNIME (Konstanz Information Miner) and Continuum Analytics, which all cover the same bases as Magellan.

In pure-machine learning terms, Amazon Web Services, Google and Microsoft are attracting developers to their cloud-based development platforms and APIs. Lastly, it's worth pointing out that there are at least 20 pure plays touting some type of advanced analytics offering with machine-learning smarts built into it, which will vie with Magellan depending on the use case and audience.

SWOT Analysis