At Staffing Advisors, we are process people. We rely on data and research to inform our choices. We build workflows and test and tweak until they are just right. And then we keep adjusting because nothing stays “just right” for long. That’s how we help clients address their most challenging hiring problems with predictable accuracy and reliable timelines while running with a small team. Now we’re using generative AI and process automation to optimize our workflows and reduce repetitive tasks, leaving more time to meet our clients’ needs.

We Don’t Use AI for Decision Making

Before we dive in, we want to be clear about how we use these tools for recruiting—and how we don’t.

We use AI capabilities in tools like LinkedIn Recruiter to add to our custom sourcing approach (although the yield from AI is typically less than 5% of the total candidates we source) and SourceWhale to automate scheduling for initial outreach. But our use of AI for recruiting stops there.

We don’t use AI for anything related to evaluating candidates. We don’t use it to review, screen, or analyze candidates’ resumes, online profiles, or responses to questions and writing prompts. Using AI to sort and rank candidates would undermine our skills-based, inclusive hiring approach. There is too much risk for AI to perpetuate the decades-old hiring biases we are working against. And it can’t grasp the nuances required for good hiring decisions.

AI and Automation: Case Studies in Productivity

Here’s the fun part (if you love efficiency like we do). Using process automation and AI, we revamped workflows for two time-consuming client deliverables and changed the way we record and distill key information from calls.

Candidate Calibration: Process Automation

Here’s a quick explanation of what the candidate calibration is. After the kickoff call for a new search, our sourcing team identifies 10-15 sample candidate profiles to gauge whether our vetting criteria are on the right track. To create the client deliverable, another team member anonymizes the profiles, removes irrelevant data, and compiles them. With more formatting headaches than seem possible for a relatively simple document, it used to take 3-4 hours just to create.

There had to be a better way.

We did some research. We experimented with different tools. And we created a new workflow. Our sourcing team still identifies and analyzes the profiles (the critical thinking part), but we reduced the rest (the copy, paste, and format part) to less than an hour and just a handful of clicks. Here’s how:

  • Using a data scraping tool, export the information from online profiles into a JSON file and convert it into an Excel spreadsheet.
  • Run an Excel macro to organize and remove irrelevant information.
  • Merge the data from Excel into a preformatted template, proofread it, and export it to share.

No more frustration with formatting. No more copying and pasting. And we can shift our focus to the more complex aspects of what our clients need. Less keyboarding, more thinking.

Reference Check Summaries: Using ChatGPT

Checking references is a valuable service we offer clients. We conduct a 15-20 minute call with every reference (for every finalist, for every search) and provide summaries for the client. Not including the call, this can take anywhere from 1-2 hours. With multiple references for each candidate, that adds up.

Creating summaries of the calls while maintaining the tone of the conversation takes time, but not a lot of judgment—a great use case for automation and AI. We reduced that time to about 20 minutes. Here’s how.

  • We record (with permission) reference calls using Teams and export the transcript.
  • To protect privacy, we run a Word macro to replace any personal identifiers, such as names or specific companies, with placeholders. We also turn off the “train the model for everyone” setting in ChatGPT.
  • A team member reviews the transcript for key points and customizes a set of prompts to guide the AI.
  • We upload the formatted and anonymized notes to ChatGPT, use the prompts to summarize, review it for accuracy, swap out the placeholder names for the real ones, paste the output into a template, and off to the client it goes.

AI-Supported Notetaking: Fireflies.ai

Kickoff calls with clients and interviews with key stakeholders are packed with important details. It’s a heavy cognitive lift for team members to take precise notes while still engaging in the call.

We started using Fireflies.ai to record and transcribe calls (with the permission of all participants), freeing our team to listen deeply, ask good questions, and take more targeted notes. The platform has AI processing features like generating call summaries and action items, and a GPT-4 interface you can customize to generate output based solely on the transcript. The result isn’t perfect, but it often helps to cut the time it takes to review and synthesize takeaways.

Although we consistently use Fireflies for kickoffs, it isn’t the best tool for every use case. Some team members use Otter.ai for other types of calls. Others prefer to record calls in Teams. And some don’t use tech at all. We each have the flexibility to test different platforms and use what works best for our role or the specific task.

(Fireflies.ai and Otter.ai are both SOC 2 (Type 2) certified and GDPR compliant. All data remains confidential and isn’t used to train their models.)

Ready-Made Solutions Rarely Fit What We Need

Everything about our recruiting and hiring process is custom. We developed Results-Based Hiring® almost twenty years ago and have been refining it since. And that means many out-of-the-box technologies aren’t designed to support what we do. But improving our processes is integral to delivering the best client service possible. So we DIY technology solutions. We aren’t tech experts, but we know what works.

The more time we save on rote tasks, the more we can dedicate to providing tailored decision-support and responding to an ever-shifting job market. We work smarter so our clients can hire better. AI and process automation are helping us along the way. (Read more about our privacy and technology practices here.)