SAS study tour 2024, North Carolina - day 1

SAS hosted a small group of press and analysts at its Cary, North Carolina headquarters this season, a location that sits adjacent to the Raleigh-Durham US East Coast technology hub.

The Computer Weekly Developer Network attended dive deep sessions focused on the SAS enterprise data and AI platform, tools and architecture with sessions on new advances in generative artificial intelligence (and pre-existing predictive and causal AI too of course), deep dive data analytics and more.

With a customer base spanning all industry verticals but with particular strength in insurance, healthcare, finance manufacturing and the government sector, SAS draws its name from Statistical Analysis System, the name of the software package that the organisation’s early developers created nearly half a century ago.

Welcome to SAS

Once known as the SAS Institute, the organisation boasts a pleasantly landscaped campus in the North Carolina woodlands. Welcoming us with an informal ‘get to know the company culture’ approach was Jay Upchurch, SAS CIO, a local native to the area. Talking about how SAS has evolved from being a data analytics company with a track record in statistical analysis to now being an AI tooling, modelling and management company, Upchurch thinks that the impact of generative AI has been ‘something of a sledgehammer’ as he put it.

“But sledgehammer or not, now is the time to really work out how to use AI for productive insights and outcomes in modern workplace environments – particularly in industry-specific use cases,” said Upchurch. “Today we’re at a point where CIOs have to decide whether the application of AI will deliver identifiable business value in real terms. This is still the prototyping phase of AI in many ways, but there comes a cut-off point now where organisations have to balance the application of intelligence and automation technologies in practical terms.”

Upchurch noted that key positive payoff areas for AI include applications and data services designed to boost employee productivity and, crucially, those that serve prompt-based software engineering (such as coding co-pilots). He says that his team is now focused on ‘working to get the enterprise value’ out of generative AI.

“Our data experience is really our crown jewel,” said Upchurch. “Just because SAS has been around for a long time does not make us a legacy software company, this is a software company with a legacy and there’s a key difference.”

Data where it resides

The SAS data analytics philosophy today has key lynchpins in data access and data management. The company says it knows that organisations want to be able to leverage analytics services on their data where it resides, without having to necessarily move it and without a firm’s software engineering team having to ‘prepare’ it (i.e. reformat it, parse it, or change it in some way) for ingestion and onward analytics. That way, data scientists can actually get on with what they enjoy doing, which is modelling and modernising systems for business value. It’s what Upchurch called the ‘sport part’ of data analytics.

What all that brings us to is the evolution of SAS which has seen the company progress through what could be defined as five stages:

  1. Language (the SAS statistical analysis language).
  2. Tools
  3. Platform
  4. Industry-specific solutions
  5. Models

Taking over the sessions on day one, Bryan Harris, CTO at SAS detailed the state of the company’s architectural stance. He believes that SAS today has one of the most advanced CI/CD pipelines that there is, plain and simple. The technology can be compiled once and run anywhere (e.g. on AWS, Google Cloud Platform, Microsoft Azure) and across on-premises installations at a variety of levels. It also supports enterprise open technologies including Red Hat OpenShift.

Data + AI + Decisions + Outcomes = Learning Rate

The above-noted ‘formula’ is Harris’ way of expressing how the SAS AI platform – a technology known as SAS Viya – is capable of helping businesses to accelerate their smart decision-making.

“The learning rate of an organisation (let’s choose life sciences as an example) would be a measure of how fast it could achieve drug discovery, or in banking, this might be measured by fraud detection. In retail, it would be a measure of supply chain efficiencies achieved, or in government, it would correspond to measuring the positive effect of tax dollars spent and reducing waste of public resources, We can apply AI at different levels across industry verticals,” said Harris.

The primary components of the SAS Viya platform include:

  • Information governance
  • Studio
  • Visual Analytics
  • Model Studio
  • Model Manager
  • Intelligent Decisioning

“SAS Model Studio is a pretty compelling tool. This technology can host a ‘model tournament’ so that we take a target variable and data set in a scenario where – let’s say – a business wants to analyse at what point customer attrition might occur,” explained CTO Harris. “The model tournament can then run thousands of models with hyper-parameter tuning in each case to find out which model performs best for the use case in hand to make the most accurate prediction.”

He notes that today, in the modern AI consumption market, there are builders of AI, buyers of AI… and now, subscribers of AI models who want a more compartmentalised and packaged approach to AI consumption.

SAS Viya Workbench

Harris centralised much of his presentation around the company’s flagship SAS Viya Workbench technology. As noted here on Computer Weekly, Viya Workbench is targeted at developers and AI modellers and works as a self-service, on-demand compute environment for conducting data preparation, exploratory data analysis and developing analytical and machine learning models.

Viya Workbench allows developers and modellers to work in the language of their choice (that choice is initially SAS and Python) with R slated to become available by the end of 2024. Engineered with a flexible interface, Viya Workbench offers two development environment options – Jupyter Notebook/JupyterLab and Visual Studio Code.

SAS company culture

Having the chance to explore company culture and step inside a software organisation of this size is always insightful. SAS was founded by software developer and entrepreneur Jim Goodnight, a man who still regularly codes and contributes to the company’s platform evolution. SAS staff all respectfully refer to the CEO as Dr. Goodnight and his status within the firm, the local township of Cary and the wider Raleigh-Durham area is unmatched.

SAS employees enjoy a warm company culture with individual offices for private work, on-site childcare facilities for parents, a resident doctor and healthcare staff and even a hairdressing salon. There’s even a southern Carolina drawl in the accent around these parts, it’s a pleasant place to be.

 

The SAS AI & data developer lifecycle.

CIO
Security
Networking
Data Center
Data Management
Close