Wolfpacks and Minesweepers: The Real-World Inspiration for Veritone Hire’s AI Algorithms


  • As part of our mission to make AI technology more transparent in the industry, we’ll highlight the key algorithms in our Veritone Hire products, revealing what inspired them and how they work. 
  • Veritone Hire algorithms leverage bio-mimicry and analogy to solve complex problems in the job advertising space. 
  • With AI, teams can better optimize campaigns at the bid or job level, maximize the budget for volume hiring, and much more. 

In the rapidly changing talent acquisition landscape, artificial intelligence (AI) has emerged as a critical tool to empower hiring professionals to do their jobs more effectively and help them source the best talent for the right job. However, part of the industry’s challenge regarding AI is a lack of transparency in how the technology works. To enable the full potential of AI, Veritone believes in the safe and ethical use of the technology. That’s why we created our AI for Good standards with four key pillars: transparency, trust, security and compliance, and empowerment. 

Veritone wants to share how our AI technology works to make the industry more transparent. In doing so, we aim to set the standard for other companies in the industry to pop the hood and showcase how AI technology works behind the scenes. Transparency is how the industry will build trust in the populace concerned about the use of the technology. 

In this blog, we will share the inspiration and unique challenges that played into developing our AI algorithms for the programmatic platform pandoIQ, a Pandologic product. Pandologic is a brand that now makes up the Veritone HR solutions portfolio. These algorithms are inspired by the natural world and utilize cutting-edge technology, transforming how job ads are bought, placed, and optimized.

The Role of Bio-mimicry and Analogy in AI Development

Veritone Hire’s programmatic AI algorithms draw inspiration from the natural world, employing bio-mimicry and analogy to solve complex problems in programmatic advertising. By studying the behavior of animals, Veritone Hire’s algorithms have been able to replicate their strategies and apply them to the recruitment process. This innovative approach has resulted in the development of highly efficient, adaptable, and effective algorithms for attracting top talent. But let’s start at the beginning—what’s bio-mimicry? 

Bio-mimicry is the concept of imitating natural processes and behaviors in designing and developing artificial systems. Veritone Hire’s bio-mimicry algorithms are inspired by various animal behaviors to optimize programmatic advertising campaigns.

For example, the Flock algorithm is one of the most notable bio-mimicry algorithms used by Veritone Hire. This algorithm is inspired by the flocking behavior of birds, where individual birds move in a coordinated manner to navigate through their environment. The Flock algorithm applies this concept to the recruitment process, allowing job ads to be strategically placed and balanced to attract the right candidates. There are three algorithms we’ll discuss in this category, including the Flock algorithm and two others called Wolfpack and STEM. 

Flock Algorithm: Balancing Job Opportunities

Drawn from Craig Reynolds’ Boids artificial life program developed in 1986, Flock was our first biomimicry algorithm. Basically, the way it’s designed is based on how a flock of birds navigates from point A to point B. It functions with three basic rules: don’t collide, try to find the average position according to your neighbors, and move in the average direction others are flying. So, how does this work with pragmatic advertising? 

By following these three rules, you can create a program that remains organized no matter the effects placed on it (you’ve probably seen a flock of birds disperse for various reasons only to come back together). With programmatic advertising, you’re trying to balance the budget against the jobs you’re trying to fill and optimize the best way you can against goals. 

Let’s say you have ten jobs, and each must achieve a performance baseline. While a flock of birds coordinates geographically, the algorithm coordinates based on spend, clicks, and the number of applicants. 

As a result, it ensures that job opportunities are distributed evenly across various platforms and channels. Like birds in a flock, the algorithm analyzes data to determine the optimal placement of job ads. It’s target agnostic as its performance oriented. This approach maximizes the visibility of job ads and increases the chances of attracting qualified candidates.

The Flock algorithm was inspired by the Boids algorithm, an artificial life simulation that tried to replicate the behavior of birds. Veritone Hire adapted this algorithm to the recruitment industry, utilizing its power to optimize job ad placement and enhance programmatic advertising campaigns’ overall effectiveness.

However, as we grew, we took on a new client that needed to conduct high-volume hiring. They needed to bring a lot of candidates in as cheaply as possible. Flock was not a good fit for this challenge, and therefore, we created the Wolfpack algorithm. 

Wolfpack Algorithm: Evolution in Response to High-volume Hiring 

Inspired by the pack hunting strategies of wolves, Veritone Hire’s Wolfpack algorithm is designed to maximize applicant volume as cheaply as possible. Wolfpacks hunt to get as much prey as possible and spread that throughout the pack. In other words, they hunt as much as they can. 

Within a pack, you have different functioning agents. The Alpha is the leader and gives orders. The Betas try to provide an alternative to the Alpha and may even overthrow the Alpha. Gammas just do what they are told, and Omegas are the ones who aren’t part of the actual hunt (the old and sick, for example). 

The basic function of the Wolfpack is to find the optimal bidding strategy for a particular job through a particular vendor. The Alpha is the best bidding strategy, commanding the Gammas to bid according to that strategy. The Beta provides an alternative strategy and tries to become the new Alpha. The Omegas are the ones “sitting out” as commanded by the Alpha to help regulate over-pacing with the budget.

This role differentiation allows recruiters to allocate their resources effectively, focusing on the most important job ads while maintaining a presence for other positions. By assigning specific roles to job ads, the Wolfpack algorithm ensures that each ad receives attention and resources and provides the highest volume of applicants. It’s a more advanced and robust algorithm, using reinforcement techniques and obtaining feedback from our actions to help you optimize the campaign according to this unique use case. 

While the Wolfpack algorithm does adapt to optimize, another challenge clients faced was greater customization capabilities for objective-based campaigns.

STEM: Adapting to Objective-Based Campaigns

Veritone Hire’s STEM algorithm takes a unique approach to programmatic advertising by focusing on objective-based campaigns. STEM, “Strategic Targeting and Efficient Management,” allows recruiters to customize their campaigns based on specific goals, such as cost-per-applicant (CPA), balance, or performance.

Like a stem cell that can turn into any type of cell, the STEM algorithm is highly adaptable and can be customized to meet the specific needs of different campaigns. Recruiters can define their campaign objectives and preferences, and the STEM algorithm will dynamically adjust the bidding and targeting strategies to achieve those goals.

The STEM algorithm starts with historical data and takes a forward-looking approach, using real-time data and predictive analytics to optimize campaign performance. This adaptability allows recruiters to stay agile and respond to changing market conditions, ensuring the success of their recruitment campaigns.

Now that we’ve covered our biomimicry algorithms, we will discuss algorithms based on real-world analogies. 

Minesweeper: Predictive Analysis for Efficient Spending

Many of us remember the logic puzzle game Minesweeper. The goal was to clear the board of squares without detonating any mines with help from clues in the form of numbers detailing how many neighboring mines there were in the area.

Taking that analogy, the Minesweeper algorithm is designed to avoid explosive or unproductive job cases by applying predictive analysis to manage spending to avoid budget spikes. This algorithm helps recruiters optimize their budget allocation and ensure a more efficient use of resources.

It does this by analyzing data and applying predictive analysis techniques to forecast the performance and cost of job ads. By understanding the potential return on investment (ROI) of different job cases, recruiters can make informed decisions about budget allocation and spending.

The algorithm identifies potential “mines” or job cases that may result in excessive spending or low ROI. By avoiding these “mines” and focusing resources on the most promising job cases, recruiters can optimize their budget allocation and maximize the efficiency of their recruitment campaigns.

One of the key challenges in programmatic advertising is the lack of real-time cut-off. Say you have a $100 budget. You wait until it shows that you’ve spent all $100 and decide to stop the campaign. But if you wait until this point and try to stop the campaign, you’ll spend more than $100 because it’s not an instant, real-time stop. The Minesweeper algorithm addresses this challenge by providing recruiters with real-time insights and recommendations based on predictive analysis.

Another challenge we faced, as with any technology, was identifying bugs as issues in the system before they became a greater problem. To detect these issues, we developed another algorithm based on sonar. 

Sonar: Detecting Anomalies in the Platform

Veritone Hire’s Sonar algorithm is crucial in detecting anomalies within the programmatic advertising platform. Sonar identifies any irregularities or unusual behavior that may indicate potential issues or opportunities by monitoring data and analyzing patterns.

For example, a job posting vendor sends us daily files on the traffic they see on our platform. Suddenly, they started sending empty files, which meant they stopped counting clicks somewhere along the line. The Sonar algorithm automates anomaly detection, allowing recruiters to identify and address issues. 

The process still functions with a “person on the loop,” but eventually, this process will include a fully automated component. By leveraging the power of AI and machine learning, Sonar provides accurate and timely insights, enabling recruiters to take proactive measures and optimize their programmatic advertising campaigns.

Carousel: High-level Job Distribution Decision-making

For the final algorithm we’ll cover, we needed a way to facilitate high-level job distribution decision-making—enter the Carousel algorithm. Much like how a carousel spins, the algorithm helps recruiters serve the best permutation of jobs at any given time. Whereas Wolfpack is at the bid level, the carousel is at the job level. It determines which jobs are going to get a seat on the carousel.

Why is this important? Let’s say you have a campaign budget of $10K. You don’t want to spend everything on the first day or week. That means not all jobs can go out and be distributed simultaneously. You need the best combination of jobs to deliver the best results for these constraints at any given moment. 

The campaign-level goals set the constraints. You want to spend a specific amount of money at a specific pace against your CPA goal. Normally, you would have to manually pull back if you’re spending too fast or select easier jobs to fill to reduce to CPA. It’s a balancing act between campaign goals and putting each job where it should be at any given time. 

By balancing campaign goals with individual job distribution, the Carousel algorithm enables recruiters to optimize their recruitment efforts and achieve better results. This approach maximizes the impact of job ads and increases the chances of attracting top talent.

AI That Empowers Great People to Find More Great People 

Veritone Hire’s AI algorithms have enhanced the teams’ capabilities to find the best people for the right job at the right price. With AI enhancing the human who has the domain expertise and know-how, teams can transform how they buy, place, and optimize job ads. This results in a more efficient and effective recruitment process, attracting top talent and improving the overall success of programmatic advertising campaigns.

As the recruitment industry continues to evolve, Veritone Hire’s AI algorithms will play a pivotal role in driving innovation and shaping the future of programmatic advertising. Through our programming advertising platform and other AI-based offerings, recruiters can stay ahead of the competition and achieve greater success in their talent acquisition efforts.

Learn More about Veritone Hire Solutions