From recruitment to analyzing developer productivity and security, AI can speed and improve decision making.
CIOs are working overtime to determine how to leverage AI. Angelic Gibson, CIO at AvidXchange, a leading provider of accounts payable (AP) automation software and payment solutions for middle market businesses and their suppliers, describes how she is using AI for everything from recruitment to managing the microservices in their SaaS platform.
Bob Scheier: How is AI helping you with talent recruitment and acquisition?
Angelic Gibson: We are using AI capabilities in the Greenhouse hiring platform in partnership with our talent recruiters to comb through resumes for the critical skills we need, both in technical skills and behavior. We use AI to scour hundreds of resumes to identify, streamline the down selection to the right people and accelerate our time to fill and our retention of people.
How do you train the tool to scan for the technical skills and behaviors you need?
The talent acquisition team does the training of the system, indexing the quality of the resumes it provides. Over time, it learns based on what we identify as high quality.
How long did it take to train?
The Greenhouse platform was implemented in mid-December 2023. We conducted personalized trainings for our hiring managers and for our basic users to ensure a seamless transition between systems. Greenhouse is extremely intuitive and requires little ramp up time, and we're already seeing a positive response from our hiring managers, candidates, and teammates. We estimate that with this system we may be able to reduce our time to hire by as much as 20% while increasing quality and we are confident that these remarkable figures will have a significant impact on the organization.
How is AI trained to identify an individual’s technical skills and behavior in order to determine their suitability for a positive cultural fit?
We have three critical mindsets. These are, first, being “Customer Obsessed” – our customers are top priority enterprise-wide. Second is having a “Growth Mindset;” does this candidate want to learn, to grow, or think about all the obstacles that prevent them from finding solutions for something? The third is the candidate’s ability to connect with others in the workplace or as we call it, “Connected as People.” Does this person just work within their “pod,” or do they often reach across formal lines into the business so they can really get to know stakeholders and learn how to work with them to solve problems?
AI can glean insights into a candidate from their use of social media, to get a fuller view of a person. We also get a lot of insights from the Azure Cloud Collaboration Center, identifying who people are collaborating with, and are they meaningful collaborations, and where there could be missing links if we have, for example, a team of people who should be collaborating with sales, but we’re not seeing log data to indicate that, that could mean an issue we need to address through recruitment.
AI can analyze calendaring and chat activity on Microsoft Teams to tell us who our teams are engaging with. It isn’t bullet proof, but it does give us an idea of where we might need to be learning and building higher connectivity. We also have a portfolio of skills we need on our team, so we can assess, through online training, how people are progressing.
Can you provide an example of those skills?
As a SaaS firm, we need a modern, nimble, micro services architecture to deliver the services our customers need. Not all engineers know how to contribute to that. We use AI to inform which patterns in our infrastructure provide which reusable services, as well as their quality and how quickly design teams deliver them. We can start to understand how productive our various teams are, and if we look out two quarters, will we have the type of capabilities we need. We also use AI to track signals from our platform for insights into issues, such as a lack of computing power, that might not be a problem today but could be over time.
Related article: Taming the AI - Anxious Inertia - in AI Decision Making By Kenneth Corriveau |
Where else are you using AI?
First and foremost is cybersecurity. We use AI to find insights, trends, into what the different threat actors are doing out in the world, and how could they become a problem for us, and to ensure we have a security procedure to mitigate future risks. For the help desk we’re using AI to serve up answers such as “How do I reset my password?” or how to perform common functions in platforms such as Salesforce or NetSuite. And we’re using AI chatbots to interact with our customers and to enable self-service, as well as to develop systems that can correct themselves.
In software engineering Microsoft Copilot allows us to take the mundane, repetitive tasks out of the software engineering life cycle, allowing development to proceed more quickly. It also increases productivity by inspecting for potential code defects, increasing quality and speed to market. We’re also using it in marketing, to increase the efficiency of our marketing communications, to build a stronger pipeline.
Can you quantify any of the benefits?
Some folks are quoting 70 percent productivity improvements in software engineering. We’re seeing 30 percent improvement, which is still pretty good. It’s still early and we’re treading cautiously. We want to protect our critical data. We have identified the lower risk workloads, where we can get increase productivity without increasing risk.
How are you using AI in your own work?
One is for quick first drafts of job descriptions. We have a solid library of job descriptions, but there’s a huge opportunity to use AI to give them a face lift, with a cultural component of excitement. Another is creating RFPs for vendors, which can be highly complex and typically take 40-60 hours to create. AI can create a baseline RFP we can then tailor to our needs.
How does AI help with job descriptions?
With the changing technology landscape, new roles are being created nearly consistently. AI itself is creating a lot of new roles. It can tell you how other companies are shaping these roles and provide a foundation to get you started more quickly in developing these descriptions for yourself.
How can AI help you make better decisions more quickly?
One example is our Invoice Accelerator, which can provide advance payments to suppliers on our network more quickly than their customers pay. We can look at all the data on the network and be very precise in who would offer such invoice acceleration to. Another area is the millions of invoices we receive, in varying formats. We partnered with Microsoft to develop the Azure Forms Recognizer, which is more precise than OCR because it analyses individual pixels on the page. This improved the quality of the data we pull from the documents, reduced our exception rates, and increased the quality and output of the critical payment data we need. It improves the quality of the business workflow, for faster decisions.
What advice would you have for others?
You have to be really good at the basics of data quality before implementing AI. Data protection and stewardship across the enterprise is table stakes, and where you need to start.
Before joining AvidXchange in 2018, Angelic was senior vice president, information technology at Technekes. During her two decades of IT experience Angelic gained deep experience in enterprise technology systems, as well as management practices supporting databases, networks, telecommunications, and infrastructure
Add a Comment