Teradata faces a challenge – how to grow a company when a key resource — data scientists and top data analysts — is limited.

Victor Lund, who was named president and CEO of Teradata a year and a half ago, said the company is focused on the top 500 corporations around the world.

“We limit ourselves to 500 companies because we can only deploy X number of people,” explained Lund. Focusing on the largest companies makes sense with Teradata’s capabilities. “We scale better than anyone else, and in real-time analytics that matters. We look at long-term relationships with our customers.”

The company is training its sales force in how to interact with the business users without threatening the IT buyers who have been the company’s usual point of contact. As data, organizing it and making use of it become a topic for the CEO and in many cases the board, the Teradata sales force has to adapt.

“We need to understand how to interface with the business but avoid threatening IT so they don’t think we are driving over them.”

At the company’s annual Partners conference, senior executives briefed reporters and analysts on the company and the growing understanding of the importance of data and analytics.

Data can be used for asset optimization, they said, and this refers to physical assets, not financial. As an example they pointed to Maersk, the shipping company, which has millions of containers. Sensors can show where the containers are, monitor them for changes in temperature or air pressure, and feed a database with all the information — real-world example of the internet of things (IoT). With Teradata firms can analyze anything, display anywhere, buy any way, including pay as you go, and move the data anytime from in-house to cloud at no cost.

Teradata is the only company to use the same code across all deployments, whether AWS, Azure, or on-premise hardware.

The company said its solution is faster and cheaper than a leading competitor’s. One million queries on the competition costs $605,000 and takes nine months, while on Teradata it can run in 10 minutes for $60.

Companies still have a way to go in their use of data, said Rick Farnell, senior vice president at Think Big Analytics, which Teradata acquired in 2014. Fifty-two percent of firms in a survey said they are at a basic level of analytics maturity and 41% of CIOs said data scientists, business intelligence and analytics are among their top skills gaps. Nonetheless, 80 percent are investing in artificial intelligence even though 34% said they don’t have the right in-house skills.

To help companies fill the gap in skills, and to leverage the people skills at Teradata, the company is building accelerators composed of best practices, code, IP, and proven design patterns to help accelerate deployment of AI solutions and ensure quick ROI. They include AnalyticOps Accelerator, which provides an end-to-end framework to facilitate the generation, validation, deployment, and management of deep learning models at scale. This accelerator is available now.

The Financial Crimes Accelerator uses deep learning to detect patterns across retail banking products and channels such as credit card, debit card, online, branch banking, ATM, wire transfer, and call centers. It continuously monitors and thwarts fraudulent schemes used by criminal actors to exploit the system. This accelerator is being deployed in Q4, and will be available more broadly in the first quarter of 2018.

“We’re building capabilities into the product,” said Oliver Ratzesberger, EVP and chief product officer of Teradata.

A recurrent theme at Teradata is around analytics at enterprise scale, not in a lab or running as a proof of concept.

“Companies need to embrace analytics at scale,” said Ratzesberger. “Prepare to disrupt, put outcomes into production at scale. Companies struggle to derive the outcomes they aim for.”

Part of the trouble is finding clean data.

“It comes down to a single source of the truth,” said Rick Farnell, senior vice president, Think Big Analytics. “That’s difficult if you have multiple organizations [with their own data silos] out there. You need to base everything off the right data.”