Skip to content
English
Customer portal
Submit a request
  • There are no suggestions because the search field is empty.

PeopleInsight AI FAQ

What is PeopleInsight AI Assistant (PIA)?

 

What is PeopleInsight AI Assistant (PIA)? 

PeopleInsight AI Assistant (PIA) is an advanced AI-powered companion for the PeopleInsight platform that analyzes data within PeopleInsight and delivers actionable insights to users, saving time, simplifying data exploration, and enabling strategic decision-making. 

 

What types of insights does the solution provide?  

PIA offers 2 distinct types of insights: 

  1. Descriptive Insights: PIA analyzes and summarizes data to highlight trends, patterns, and flag exceptions where KPIs exceed established threshold values so users can quickly get a comprehensive view of organizational performance.  
  2. Diagnostic Insights: PIA identifies the root-cause of exceptions in KPIs by analyzing the data and helping users understand the ‘why’ behind the exceptions. 

 

What are the benefits this solution offers? 

PIA offers several benefits including: 

  • Save Time with Instant Access to Insights: Skip the hassle of navigating multiple dashboards to get answers to your questions—simply ask questions in natural language and get precise answers in real time. 
  • Inform your HR strategy with automated, objective analytics you can trust, uncovering the "what," "why," and "what's next” -  no data science expertise required 
  • Simplify Complex Data: Arm stakeholders with answers, without the need for technical or data science skills, making sophisticated insights accessible to everyone. 

 

How can I access the solution? 

  • PIA Overlay - PIA will appear as an overlay on top of your existing dashboards to surface descriptive, diagnostic and predictive insights 

 

What AI technologies power this solution? 

  • Descriptive Insights: We utilize statistical analysis techniques to identify trends in the data and evaluate if key metrics are beyond expected thresholds. We leverage LLMs to provide user-friendly summaries of the insights we find in the data. 
  • Diagnostic Insights: We use a variety of machine learning and generative AI techniques to power our services.  With machine learning, we apply anomaly detection and feature importance extraction techniques to understand and explain root-causes of increase/decrease in KPI metrics.  We leverage LLMs to provide user-friendly summaries of the machine learning analysis. 

 
How does this solution ensure data privacy and security? 

  • Protecting our clients data is of utmost importance for PeopleInsight. To ensure that all customer data is safe and secure, PeopleInsight maintains SOC2 type 2 certification. 
  • All data in transit and at rest is encrypted using industry-leading security protocols, ensuring unauthorized access is prevented at every stage of data processing. 
  • In addition to the above, we ensure that AI models and services only access and process data that customers have explicitly authorized, upholding strict consent protocols to guarantee transparency and maintain trust. 
  • The instances of the AI/ML models we use through AWS services such as Bedrock and Sagemaker are contained within the PeopleInsight environment and the data processed by these models is never shared with model providers or used to train the generalized foundation models.  
  • Our AI services are designed to operate without utilizing Personally Identifiable Information (PII), ensuring that sensitive data is never processed or exposed, thereby protecting user privacy and reducing compliance risks 
  • Our AI solution integrates privacy by design ensuring that data protection and
    privacy rights are fundamental in our system architecture and operational
    practices:
    • Data minimization – we only ingest and process data that is needed to
      produce insights.
    • Data anonymization - data is always anonymized by aggregating above
      the individual level.
    • Data retention – data is removed as soon as it is removed from the client
      dataset and is not kept longer than necessary to execute the data refresh.
    • User Control – data access is controlled at the user level and restricted to
      only what is needed for the user to execute their functions.

 

How can we be sure of the accuracy of the insights? 

  • Accuracy of insights generated by the AI services depends on having good, clean data to begin with.  
  • In PeopleInsight by HireRoad, data quality and accuracy are of the utmost importance.  Our data extraction, transformation and loading processes from customer source systems are built based on a framework of robust data audits, which ensures we have good clean data.  
  • On top of the good clean data, our AI services leverage robust, industry-standard algorithms and models that have been tested and validated against real-world data. These models are designed to deliver accurate and reliable insights based on proven data science principles.
  • Before being rolled out to customers, our AI solutions undergo rigorous testing to
    ensure that they perform accurately. This helps us ensure that the insights we
    provide are reliable and accurate.

 

What technology platform do you use for your AI technology?

  • Our AI architecture leverages the modern, secure, and scalable capabilities of
    AWS to drive innovation and deliver exceptional performance. With enterprise-grade security, global availability, and cutting-edge AI tools, AWS empowers us
    to create reliable, sustainable, and high-performing solutions that scale with
    ease.
  • Our architecture is currently built into AWS using infrastructure-as-Code (IaC)
    methodology to ensure service stability and performance. IaC also allows us to
    easily build new architecture for features or migrate to other cloud providers.
  • We leverage several data science features within AWS ecosystem including AWS
    SageMaker to train and optimize model accuracy and AWS Bedrock for LLM
    management.
  • To minimize energy consumption and improve service speeds, we use services
    such as AWS Lambda, Glue, and Athena for processing data on an automated
    schedule.

 

Is there an additional cost to using this solution?

  • No. For PeopleInsight by HireRoad, we believe in delivering exceptional value
    without compromise. That’s why our advanced AI features, including the
    PeopleInsight AI Assistant (PIA), are available at no additional cost. We’re
    committed to enhancing the overall solution for our users, empowering them
    with powerful analytics and conversational AI to drive better decisions. While we
    may explore premium/paid options in the future, our current focus is on
    providing an unparalleled experience that sets a new standard in people
    analytics.

 

How are we ensuring responsible and transparent AI?

We are committed to ensuring responsible use of AI by adhering to below
principles that are considered as key tenets of responsible AI.

  • Fairness and Inclusivity: We ensure equitable outcomes by auditing datasets,
    addressing biases, and incorporating user feedback into model improvements.
  • Transparency and Explainability: Our AI models and decisions are well-documented, with plans to share accessible insights with users for greater
    understanding.
  • Privacy and Security: We prioritize data protection with SOC-2 certified
    systems, encryption protocols, and AI services that exclude PII and process only
    customer-authorized data.
  • Accountability and Reliability: Continuous monitoring, feedback loops, and
    robust testing ensure our AI systems operate safely, reliably, and as intended.
  • Environmental Responsibility: Our AI systems are optimized for energy
    efficiency, scaling dynamically based on usage to minimize environmental
    impact.

 

Can users influence/train the AI the types of insights they want to see?

  • No. Training the AI model to generate specific types of insights is a process that
    is owned and undertaken by the PeopleInsight team.
  • We believe in providing powerful AI capabilities without overwhelming our users.
    Instead of requiring users to train their own models or deal with complex
    configurations, we’ve designed our solution to deliver immediate, actionable
    insights right out of the box.
  • The solution will have feedback loops using which users can influence the
    behavior of the AI models by providing feedback on the insights/answers
    generated by the AI models. Positive feedback will further reinforce the AI
    model and negative feedback will be used to retrain/refine the models.