We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 – August 3. Join AI and data leaders for insightful talks and exciting networking opportunities. Learn More
AI is becoming increasingly mainstream – very quickly, and across a multitude of areas and applications.
According to Jack Vernon, senior research analyst, European AI Systems, for IDC, 69% of organizations are either using AI already or plan to in the next 24 months.
“But even more,” Vernon said, “AI has clearly moved from the experimental phase to mission critical, with businesses realizing real value, from improved growth and revenue, to cost reduction, to operational efficiency. Now more than ever, businesses need an AI platform that is adaptive, shifting and adjusting to even the most unpredictable market conditions.”
DataRobot, a widely deployed global AI platform, says that it can meet demands that aren’t only rapidly increasing but becoming ever more sophisticated.
Navigating the unpredictable
The Boston-based company has released Cloud 8.0, which is designed to help businesses navigate unpredictable events and market conditions. The platform’s continuously optimized machine learning (ML) models deliver top performance, connect with broad ranges of data and generate complete and accurate predictions, according to Nenshad Bardoliwalla, chief product officer.
“AI is the technology that companies must adopt in order to handle the unpredictable world we live in,” said Bardoliwalla. “The only way to navigate twists and turns in the market and geopolitical challenges, is to be able to smartly harvest your data. Data is a reflection of the real-world entities that matter to a business.”
Continuous AI is a “very powerful idea,” Bardoliwalla said. It allows companies to create multiple MLOps strategies – standing for machine learning (ML) and devops – to constantly refresh production models. This can be done at scheduled times or if a “data drift” or drop in accuracy occurs. Continuous AI also uses DataRobot’s AutoML capabilities to automatically create and recommend new models to allow organizations to be prepared for any real-world situation.
“The more models you add, the more you need the automation,” Bardoliwalla said. “What if we could actually automate the entire AI lifecycle? That’s why we call it continuous AI. It’s a constant process.”
Research firm Cognilytica estimates the MLOps market to reach $4 billion by 2025 – up from $350 million in 2019. Top players in MLOps include DataRobot, Dataiku, Databricks, Cloudera, and Iguazio. The biggest players Microsoft, Amazon, Google, and IBM also have some type or another of an MLOps offering (or plan to ramp one up soon).
For its part, DataRobot AI Cloud 8.0 is available to all businesses with a multi-cloud architecture and is easily deployable to public clouds and on premises in the data center and at the edge.
Notably, the DataRobot app builder has been extended with support for its Automated Time Services platform. This platform breaks time series into component parts and aggregates them into groups with automated clustering; prediction explanation and segmented modeling can be realized in no-code environments. This function allows organizations to be more resilient and meet the needs of a more fluid, unpredictable landscape.
In an example of time-series forecasting, customer 84.51°, the analytical subsidiary of retail chain Kroger, has used DataRobot capabilities to determine what products need to be in what stores, and when. (Say, an extra 200 loaves of bread on the shelves of a certain city’s store on a Friday afternoon.)
AI and ML for the nontechnical
As Bardoliwalla explained, the goal is to put ML-creation capabilities into people’s hands, even if they’re not technical, and allow them to use model information for real-world decision making.
As part of this, models provide explanations for their forecasting. (So to go back to bread, explaining that consumption goes up over the weekend, there’s a big weather event coming, or an anticipated increase in demand due to a long holiday weekend.)
“Technology is only as good as the proof that we have that it works,” Bardoliwalla said. “To get people who are not data scientists to use AI and ML, they have to trust it. We want to democratize and consumerize and bring AI out to a much broader audience than the coders in an organization.”
As another example, DataRobot customer Adecco Group applied MLOps to improve and increase its hiring. The initial project was gauged on the three metrics of productivity, accuracy nand interpretability, and 60-plus projects were launched using more than 3,000 models. In turn, Adecco saw a 37% reduction in the number of CVs it needed to review. As Bardoliwalla noted, models were granular and made use of ethical AI and bias metrics to reduce risk and centrally monitor and manage models with MLOps.
In another example, Oyak Cement used AI to increase its alternative fuel usage by seven times and cut nearly 2% of its total C02 emissions, thus reducing costs by $39 million.
“We’re empowering businesses to better anticipate moments of change and continuously optimize machine learning models, even those already in production, while driving new and more accurate decisions down to front line business users,” Bardoliwalla said.
With the release of Cloud 8.0, DataRobot’s Continuous AI capabilities are available in on-premises environments and leading public clouds. This helps protect against real-world changes from prolonged pandemic conditions, changing economic climates, and shifting consumer behavior. Continuous AI automatically adapts and recommends to these trends and suggests best models given historical data and current conditions.
Meanwhile, DataRobot’s library of connections has been integrated with Microsoft Active Director and Scoring Code for Snowflake. Companies now have access to an expanded model library, a complete set of pre-built integrations, and write-back capabilities to most popular cloud data stores.
As Bardoliwalla noted, from ongoing challenges in the market due to the prolonged pandemic, to unreliable support chains, to the rapidly approaching return to work, “AI has the potential to help every business manage through this unprecedented time. But your AI platform must be able to anticipate and adapt faster and more intelligently to even the most unpredictable market conditions.”
Looking ahead in continuous AI, he said, means creating ubiquitous connectivity and getting “as many hooks” into data as possible. All businesses can benefit from mathematical, repeatable, empirically based prediction and “what-if” analysis. They can raise revenues, lower costs and risks, and arm themselves with “incredible access to information” with the use of ML and AI.
“Companies that are most successful have incredible access to information and incredible ML capabilities,” Bardoliwalla said. “You can take any business that is rear looking and turn it into what might happen. They need to be able to take action to make the future what they want it to be.”
VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Learn More