Oracle is sprucing up its customer data platform, CX Unity, with a little machine learning.

The platform will now support real-time behavioral data collection and personalization capabilities through Infinity, Oracle’s digital streaming technology. Infinity captures web event, app and point-of-sale data to help brands build realistic representations of the customer journey.

That previously required repeated and rigorous A/B testing, said Rob Tarkoff, EVP and general manager of Oracle Cloud CX and Oracle Data Cloud.

“But by applying machine learning to that, you can come up with predictions and insights based on a holistic set of data, and it doesn’t require you to do one-off activities,” he said.

Marketers can use ML in this context to anticipate the best possible discount to offer at the beginning of a conversation with a customer, for example, or to reduce the number of touches required for a sale.

CX Unity launched in 2018, right around the time Salesforce and Adobe started ginning up their own CDPs. The marketing clouds have gone big on CDPs after a slow start and initial skepticism about the need for CDP technology.

But Oracle differentiates itself with what Tarkoff calls “a true one-platform story” – as opposed to “the fake news we hear from some of our competitors who only like to talk about a single platform.”

Take Salesforce, he said. Einstein, the AI layer within Salesforce’s platform, is built on Amazon Web Services; ExactTarget runs on Azure; and other parts of Salesforce are built on Salesforce’s own proprietary stack. But Oracle weaves AI into its offerings consistently across platforms, including its CDP, Tarkoff said.

“Our ambition, and a big one for Larry [Ellison], is to have a single platform orientation and to automate and engineer the front office processes as much as the backend office processes,” he said.

What that means in practice is capturing and applying signals regardless of where they come from.

Field service is a good example. Last year, Oracle introduced a Service Logistics Cloud that helps service departments collaborate, share data and manage their supply chains. Through a combination of Service Logistics and CX Unity, it would be possible to send a signal to a field technician if they need to reroute their schedule because a part they need for a job isn’t available until later in the day.

“While we do all of the marketing use cases that traditional CDPs are expected to do, like campaign suppression, personalization and real-time segmentation,” Tarkoff said, “our view is that customer intelligence expands way beyond marketing.”

The post Oracle Adds Machine Learning Capabilities To Its CDP appeared first on Artificial Intelligence.

Data science is the name of the game these days for companies that want to improve their decision making by tapping the information they are already amassing in their apps and other systems. And today, a startup called Mode Analytics, which has built a platform incorporating machine learning, business intelligence and big data analytics to help data scientists fulfill that task, is announcing $33 million in funding to continue making its platform ever more sophisticated.

Most recently, for example, the company has started to introduce tools (including SQL and Python tutorials) for less technical users, specifically those in product teams, so that they can structure queries that data scientists can subsequently execute faster and with more complete responses — important for the many follow-up questions that arise when a business intelligence process has been run. Mode claims that its tools can help produce answers to data queries in minutes.

This Series D is being led by SaaS specialist investor H.I.G. Growth Partners, with previous investors Valor Equity Partners, Foundation Capital, REV Venture Partners and Switch Ventures all participating. Valor led Mode’s Series C in February 2019, while Foundation and REV respectively led its A and B rounds.

Mode is not disclosing its valuation, but co-founder and CEO Derek Steer confirmed in an interview that it was “absolutely” an up-round.

For some context, PitchBook notes that last year its valuation was $106 million. The company now has a customer list that it says covers 52% of the Forbes 500, including Anheuser-Busch, Zillow, Lyft, Bloomberg, Capital One, VMware and Conde Nast. It says that to date it has processed 830 million query runs and 170 million notebook cell runs for 300,000 users. (Pricing is based on a freemium model, with a free “Studio” tier and Business and Enterprise tiers priced based on size and use.)

Mode has been around since 2013, when it was co-founded by Steer, Benn Stancil (Mode’s current president) and Josh Ferguson (initially the CTO and now chief architect).

Steer said the impetus for the startup came out of gaps in the market that the three had found through years of experience at other companies.

Specifically, when all three were working together at Yammer (they were early employees and stayed on after the Microsoft acquisition), they were part of a larger team building custom data analytics tools for Yammer. At the time, Steer said Yammer was paying $1 million per year to subscribe to Vertica (acquired by HP in 2011) to run it.

They saw an opportunity to build a platform that could provide similar kinds of tools — encompassing things like SQL Editors, Notebooks and reporting tools and dashboards — to a wider set of users.

“We and other companies like Facebook and Google were building analytics internally,” Steer recalled, “and we knew that the world wanted to work more like these tech companies. That’s why we started Mode.”

All the same, he added, “people were not clear exactly about what a data scientist even was.”

Indeed, Mode’s growth so far has mirrored that of the rise of data science overall, as the discipline of data science, and the business case for employing data scientists to help figure out what is “going on” beyond the day to day, getting answers by tapping all the data that’s being amassed in the process of just doing business. That means Mode’s addressable market has also been growing.

But even if the trove of potential buyers of Mode’s products has been growing, so has the opportunity overall. There has been a big swing in data science and big data analytics in the last several years, with a number of tech companies building tools to help those who are less technical “become data scientists” by introducing more intuitive interfaces like drag-and-drop features and natural language queries.

They include the likes of Sisense (which has been growing its analytics power with acquisitions like Periscope Data), Eigen (focusing on specific verticals like financial and legal queries), Looker (acquired by Google) and Tableau (acquired by Salesforce).

Mode’s approach up to now has been closer to that of another competitor, Alteryx, focusing on building tools that are still aimed primarily at helping data scientists themselves. You have any number of database tools on the market today, Steer noted, “Snowflake, Redshift, BigQuery, Databricks, take your pick.” The key now is in providing tools to those using those databases to do their work faster and better.

That pitch and the success of how it executes on it is what has given the company success both with customers and investors.

“Mode goes beyond traditional Business Intelligence by making data faster, more flexible and more customized,” said Scott Hilleboe, managing director, H.I.G. Growth Partners, in a statement. “The Mode data platform speeds up answers to complex business problems and makes the process more collaborative, so that everyone can build on the work of data analysts. We believe the company’s innovations in data analytics uniquely position it to take the lead in the Decision Science marketplace.”

Steer said that fundraising was planned long before the coronavirus outbreak to start in February, which meant that it was timed as badly as it could have been. Mode still raised what it wanted to in a couple of months — “a good raise by any standard,” he noted — even if it’s likely that the valuation suffered a bit in the process. “Pitching while the stock market is tanking was terrifying and not something I would repeat,” he added.

Given how many acquisitions there have been in this space, Steer confirmed that Mode too has been approached a number of times, but it’s staying put for now. (And no, he wouldn’t tell me who has been knocking, except to say that it’s large companies for whom analytics is an “adjacency” to bigger businesses, which is to say, the very large tech companies have approached Mode.)

“The reason we haven’t considered any acquisition offers is because there is just so much room,” Steer said. “I feel like this market is just getting started, and I would only consider an exit if I felt like we were handicapped by being on our own. But I think we have a lot more growing to do.”

The post Mode raises $33M to supercharge its analytics platform for data scientists appeared first on Artificial Intelligence.

Feature Store is a core part of next generation ML platforms that empowers data scientists to accelerate the delivery of ML applications. It enables the teams to track and share features with versioning enabled and serve features for model training, batch, and real-time predictions. Mike Del Balso from Tecton.ai and Geoff Sims from Atlassian recently spoke at Spark AI Summit 2020 Conference about the feature store driven ML development.

Del Balso talked about machine learning process shortfalls like limited predictive data, long development cycles, and painful path to production which typically involves multiple teams, lot of resources and different implementations. He spoke about Operational ML, which basically consists of applied ML solutions that drive user experiences in use cases like fraud detection, click-thru rate (CTR) prediction, recommendation, and search. Building OperationalML applications is very complex and data is at the core of that complexity. Actual ML code is a smaller portion of the overall effort compared to tasks like configuration, data collection, feature engineering, and resource management.

Features are major building blocks of any ML application but the current tooling for managing features is not where it needs to be. There is a need to automate the process of deploying and operating the feature pipelines in production, including feature engineering and feature serving.

Del Balso discussed Tecton, a data platform for machine learning applications, that automates the full operational lifecycle to make it easy for data science teams to manage features throughout the lifecycle in a typical ML process. It can be used to extract data from data sources (batch or real-time) and transform that data as feature pipelines, and organizes feature values in a Feature Store. Data platforms for ML solve critical problems like managing the sprawling and disconnected feature transformation logic, building quality training sets from messy data, and deploying to production.

ML features are highly curated data in a business, but they are some of the poorly managed assets. Since each ML model typically has hundreds, if not thousands, of features to manage, this challenge makes it difficult to scale ML efforts in organizations. He recommended that features should be managed as feature data as well as feature transformation code that’s used to generate it.

He discussed some common challenges with assembling training data like stitching multiple data pipelines together, data leakage, and delivering training data to training jobs. Data science and engieering teams also face problems when deploying their models to production and moving from a batch environment to real-time. Some of these challenges are related to infrastructure provisioning and drift & data quality monitoring. An enterprise-grade Feature Store can manage the feature training and feature serving.

Geoff from Atlassian talked about how they used the Feature Store solution to automate content categorization in one of their popular products, Jira, by automatically labeling for every issue tracked in Jira. They used the feature store to collect a large amount of events, store the features per model and update it in real time, as well as generate the features and predictions.

The post Accelerating Machine Learning Lifecycle with a Feature Store appeared first on Artificial Intelligence.

Today, Hewlett Packard Enterprise (HPE) announced its acquisition of MapR business assets. MapR began as a company focused on providing a cloud data platform. They extended their message into machine learning (ML) and artificial intelligence (AI), claiming to be good support for data sources needed in those arenas. In addition to the cloud, they focused on a container message for scalability. They had some early backing, but that backing wall pulled earlier this year.

Phil Davis, president, Hybrid IT, Hewlett Packard Enterprise, said in the press release, “MapR’s enterprise-grade file system and cloud-native storage services complement HPE’s BlueData container platform strategy and will allow us to provide a unique value proposition for customers. We are pleased to welcome MapR’s world-class team to the HPE family.”

The press release also focused on the partners in the AI/ML and analytics markets more than it did on the technologies.

What’s interesting to note is that no price was announced for the acquisition. In addition, the stated purpose of working with BlueData, another acquisition focusing on container-based software for AI/ML should make folks wonder about the purpose and benefit of the acquisition. What we have is a second acquisition in the same space, but the MapR one is of a company from with the backers withdrew funding. It is reasonable to assume that HPE acquired it for the connections into the market and not for the technology. Could they have just paid for the lead list and for the partner relations?

From a non-financial evaluation, the larger companies are continuing the fast follower strategy mentioned in my previous article, but MapR didn’t have the presence that OpenAI had, pre-acquisition. I’m sure HPE didn’t pay as much as Microsoft did, but there is still a lot left unanswered in the press release and related material. That means there is no way to evaluate what kind of sense the acquisition makes.

The post HPE Acquires MapR Assets In An Attempt To Strengthen Its Artificial Intelligence / Machine Learning Portfolio appeared first on Artificial Intelligence.