Why is it important for People leaders to effectively implement AI tooling?

AI technology can help People leaders transition from order-takers to strategic business partners – by taking on routine (and primarily reactive) activities, AI tools enable HR professionals to focus on proactive, higher-value, and more strategic activities.

It improves operational efficiency by automating time-intensive tasks – HR teams spend a disproportionate amount of time on activities that are relatively unimportant to the business—except from a legal standpoint. Tasks like researching insurance policies take hours and can be effectively handled by automated tools, freeing up meaningful time to focus on initiatives that have a greater impact on business outcomes.

Enhanced analytical capabilities unlock actionable People insights – AI enables sophisticated analysis that was previously impossible or impractical, such as predictive modeling for employee retention, skills gap analysis, and performance pattern recognition across large datasets.

How should organizations begin the transition to AI tooling? Which People operations processes are most suitable for AI integration?

Ensure you have a secure environment before experimenting with AI-powered People tools – as long as you protect sensitive HR data, established platforms like ChatGPT, Claude, or Copilot can provide immediate value through simple analytics and summarization tasks.

Start with a project that can become a 90-day quick win – choose a small pilot project that will yield measurable outcomes within a quick time frame. This approach builds momentum, boosts confidence, and demonstrates value that can facilitate larger investments.

Examples of quick-win processes include:

Summarize assessments and performance reviews across the organization – determine which managers provide effective feedback and identify managers who would benefit from additional training or support
Write faster and better job descriptions – free up time so the recruiting team can spend more time finding strong candidates
Review resumes – review inbound resumes, rank selected candidates for interviews, and/or identify potential strengths and weaknesses for each candidate

Apply your quick win to additional use cases across the business – once you’ve proven that something works in one area, expand to adjacent use cases—even if they’re solving business unit problems rather than HR problems. For example, an internal AI agent that can compare resumes might be modified to determine which conversation patterns used by SDRs have the strongest conversion rates.

What are key considerations to address before you begin implementing AI tooling?

Consideration	Key Questions	Why It Matters

Platform Architecture	• Do we need an AI-first platform or a more traditional system with integrated AI-powered pieces?	Tech stack capabilities and future scalability needs inform whether an organization should use: • General LLMs • Specialized platforms • Custom-built solutions

Tool Ecosystem	• How does the tool fit within our ecosystem? • Does it integrate with the systems and tools we use?	Success depends on a tool’s ability to get the data it needs – if your new platform can’t integrate and share data effectively with critical business systems, you risk creating information silos and operating off poorly informed insights

Complexity	• What is the complexity level of this project? • How will this use case scale? • How long will we use it? • What complexities will emerge in the next 2-3 years?	Ongoing maintenance and support needs increase with complexity – if your organization won’t be able to afford having a full-time or part-time resource who can update, refine, and maintain your AI systems as business needs evolve, you might need to choose a solution that handles future iterations for you.

Data Security & Model Bias	• How secure is this platform? • How do they negate bias?	Choose technology partners you trust – when you create your own solutions, you can control how data is used and protected. That transparency disappears when you work with other companies’ technology, so it’s crucial from an operational and legal standpoint that you choose partners wisely.

When should you use a general LLM like ChatGPT vs. a platform?

Optimal Use Case Traits for LLM vs. HR AI Platform Solutions
Factor	Fit for an LLM (e.g., ChatGPT, Claude)	Fit for an HR AI Platform (e.g., Rippling, HiBob, GEM)
Complexity and scalability	Simple, well-defined tasks	Complex, evolving requirements
Internal maintenance capabilities	Internal investment can become significant over time, but might be worth it as you better understand your use case	Requires fewer technical resources for ongoing development
Timeline and iteration requirements	Custom solutions can be created quickly and often work well when a company is just starting out	More sophisticated iterations typically transition to a platform

What are the main phases of an AI implementation project?

Use 6-month integration cycles to create realistic timelines – AI implementation should take learning curves, user training, and system complexity into account. Don’t expect immediate deployment or assume that employees will be able to use the solution without extensive training.

Phase 1: Planning – spend a total of 3 months planning from an operational and technical perspective. Approximately one month should be spent scoping use cases and complexity. Expect to spend 2 months evaluating potential vendors, attending demos, and exploring sandboxes before choosing a platform or tool.

Phase 2: Implementation and Training – it generally takes at least 3 months to get tooling set up. Begin training your audience in parallel. The nature of your audience determines whether training takes ~3 months (e.g., for smaller, HR-only audiences) or ~6 months (e.g., for larger audiences across multiple teams).

Phase 3: Conduct bias testing and model validation – AI tools can amplify implicit biases and hallucinate when they’re provided with too much data. Take an iterative approach to testing and improving your model to ensure that the system can identify and flag problematic patterns. For example, you can test whether your tool can guard against bias by feeding it purposefully biased data and confirming that it can detect problems with that data.

Phase 4: Continuous Retraining and Iteration – users require ongoing education as systems improve and change. In addition to retooling your systems, you will need to retrain power users and casual users on your solution every time you make a change.

AI Tooling Across Processes

What categories of AI tooling should you consider for talent acquisition and recruitment?

TA and Recruitment

Categories of Tooling/Vendor examples		The appropriate TA tool depends on the size of the company: • Small (50-100 people) → smaller, feature-focused tools like Dover • Medium (100-500 people) → platforms like Greenhouse, Lever, or Gem let you piece specific features/tools together • Large (500-1,500 people) → ATS’s like Ashby with advanced feature sets • Enterprise (1,500+ people) → complex, enterprise-level solutions like Workday Small businesses might find these tools especially useful: Textio – tools like Textio help managers build JDs. This is especially useful for managers who’ve never written JDs before and need guidance as they scope and describe their team roles. Gem sourcing tools – these features can be implemented independently from Gem’s full ATS offering. It is useful for growing teams that need help sourcing and messaging candidates but aren’t yet ready for a more robust system.

Implementation considerations		Understand where the tool’s tech stack roadmap is going – both your org and the tool will evolve after implementation. Ideally, this evolution will make their tool even more useful to you. Determine whether they can accommodate your anticipated future needs. For example, companies that start with Gem’s sourcing feature might eventually also adopt its CRM, and finally its full ATS model. Always expect an integration to take at least a month – assume that integration will take at least twice as long as the vendor estimates. For example, a 2-week implementation estimate might assume that someone is working on the implementation for 40 hours per week.

Operations considerations		Create playbooks for how to use the tool – any tool is only as useful as your data and processes allow them to be. For AI tools, standardized processes are especially important for prompting. Iterate playbooks based on real feedback and usage behaviors – don’t expect to be able to create the ideal playbook before you have actual usage data to test against it. Adjust your playbooks based on how users actually interact with the tool instead of trying to force ideal but unnatural behaviors.

Use cases of AI in TA and Recruitment:

Job Description Creation and Optimization – AI tools like Textio help managers, especially first-time hiring managers, build effective job descriptions by providing guidance on scoping and describing team roles with data-driven language optimization.
Candidate Sourcing and Outreach – AI-powered sourcing tools can help growing teams identify potential candidates and craft personalized messaging at scale, particularly useful for companies not ready for full ATS implementations but needing sourcing support.
Process Standardization and Automation – AI tools require and enable standardized processes, especially around prompt engineering and workflow automation, making recruitment more consistent and scalable across different company sizes.
Scalable Talent Pipeline Management – AI features integrated into platforms like Gem’s CRM and ATS offerings help companies manage candidate relationships and recruitment workflows as they grow from small sourcing needs to enterprise-level talent acquisition.
Data-Driven Recruitment Decisions – AI tools provide analytics and insights that help optimize recruitment processes based on actual usage patterns and candidate interactions rather than assumptions about ideal workflows.

What categories of AI tooling should you consider for workforce planning?

Workforce Planning
Categories of Tooling/Vendor examples	Workforce planning systems are simpler to use than TA tools – Visier and Crunchr specialize in workforce planning and are useful for companies of all sizes
Implementation considerations	Implement as soon as you have a year of data in an HRIS (Human Resources Information System) – workforce planning tools need at least one review cycle to use as a baseline for future analysis.
Operations considerations	Training and cross-functional friction are the main blockers to deriving value from workforce planning tools – Finance (rather than People) often owns workforce planning—and typically prefers using spreadsheets to more nuanced and specialized workforce planning systems that require a larger array of inputs.

Use cases of AI in workforce planning:

Workforce planning from a skills gap perspective – due to the range of inputs that progressive models require, you can identify which skills are missing from your workforce, what you can upskill internally, when you will need certain skills, and the extent to which those skills are already present in your organization.
Turnover prevention analysis for talent stabilization – understanding voluntary turnover patterns can help you identify people who are at risk—often before they are at risk. For example, you might discover that the average tenure length (controlling for different labor markets) is 1.5 years. If you categorize employees who have between 1 year, 3 months and 1 year, 9 months of tenure (controlling for labor markets) as high risk, you can focus attention from an upskilling and retention standpoint on these people.
Advanced AI-enabled capabilities – modern workforce planning tools can predict training timelines, model internal development, and compare the anticipated effectiveness of different promotion strategies (e.g., promote internally vs. hire externally).

What categories of AI tooling should you consider for employee engagement and experience?

Employee Engagement and Experience

Categories of Tooling/Vendor examples		Surveying tools to help get a pulse on employee sentiment – there are numerous survey tool options, such as Culture Amp. Conversational AI and virtual assistants – such as Leena AI, Espressive, and Moveworks offer always-on, chat-based support for employees. Employee experience and engagement platforms – like Qualtrics EX, Peakon (now part of Workday), and Humu.

Operations considerations		Choose a tool that integrates with communication platforms your team enjoys using – participation is one of the greatest hurdles in engagement surveys. The most effective tools integrate directly with Slack, Teams, etc. and don’t require separate logins. Consider survey response bias when interpreting results – employee satisfaction is always less positive than your survey says. Traditional engagement surveys typically capture only 85% of the workforce, and the missing 15% is usually the least engaged.

Use cases of AI in Employee Engagement and Experience:

Employee surveys – AI can recommend effective survey questions based on organizational context, and analyze responses to produce insights.
Chat-based support for employees – to resolve HR questions, guide employees through tasks like onboarding or benefits enrollment, and surface nudges for performance reviews or learning modules. When deployed thoughtfully, they reduce HR workload and boost employee satisfaction, but success depends on integration with core systems (HRIS, ITSM) and maintaining a tone that aligns with your company’s culture.
Mapping key moments in the employee journey – AI can provide personalized nudges or manager prompts at critical moments, improving retention and overall satisfaction. For meaningful insights, it’s essential to connect this data with other sources like attrition trends or promotion velocity, and to maintain transparency about data use.

How do you encourage a culture of adoption and enthusiastic embrace of AI tooling?

Create role or team-specific use cases – an employee should be able to understand how each relevant use case can make their life easier. People are more likely to change a process if the updated option is objectively and observably more effective.

Make it safe for employees to use tools – clearly communicate the guardrails and security measures that protect employee data. For example, if you create an L&D tool, employees shouldn’t worry that their manager will get a notification that they asked the tool how to get a promotion.

How should the People org leverage LLMs across their everyday work?

Establish usage policies and governance – establish clear protocols for using LLMs with employee data, including who has access to which environments and systems.

Teach your team about the applications of LLMs – many HR professionals don’t realize the full potential these tools represent for their day-to-day work. Regular training sessions should demonstrate creative use cases that help employees think about these tools in new ways.

Customize your trainings to different levels of employee ability – different employees will use AI tools with varying degrees of intensity, consistency, and comfort. Power users need extensive, detailed training, while casual users might benefit more from having platform fluency and an understanding of what the system should be used for.

What security frameworks should companies implement when deploying AI in people operations to protect sensitive employee information?

Store agent information in isolated server environments – AI systems that handle employee data must operate in secure environments that are separate from other organizational systems. This prevents unauthorized access to personal data (e.g., offer letters, negotiations, salaries) and can help contain potential security breaches.

Create a dedicated People IT resource – consider embedding an IT professional within the people team. This individual should understand both technical requirements and HR law/ data sensitivity and is better equipped to design secure and compliant AI solutions.

Note: IT and People teams are becoming increasingly intertwined – due to the importance of People data, some organizations are moving IT functions under the Chief People Officer to better manage sensitive data requirements.

Overall

What metrics do you use to measure successful adoption?

The most useful metrics include:

Power user usage – monitor daily usage patterns among those who use the system most. These should be your product champions; if power users are adopting the platform, it’s a good early indicator of system value and usability.
Casual user usage – track usage growth among managers, hiring managers, and executives who use AI tools less frequently. Increasing engagement from these users indicates growing confidence and perceived value.
Total hours spent in the system – successful implementation should show increasing usage over time across all user categories. If usage decreases over time, the tool has not been successfully incorporated into regular workflows.

Note: executive adoption also indicates success – if a leader has recognized the value of a tool, they are more likely to encourage their teams to adopt it and be open to extended applications in the future.

What use cases should HR users be cautious about leveraging AI for?

Generally, be wary of relying on AI without human oversight – AI is an amplifier, not an independent operator (yet). Replacing human recruiters entirely can create significant operational risks. For example, using AI to automate candidate screening calls is risky because AI:

Can get trapped in loops
Cannot answer complex questions about company culture or role specifics
Provides a sub-par candidate experience

Be very cautious in using AI tools for initiatives related to diversity, equity, and inclusion – due to the nature of implicit and explicit biases, AI tools are not ready to assist in this area. For example, Workday is currently being sued for accidentally developing an algorithm that discriminated against candidates who are over 40.

What are common pitfalls?

Overestimating abilities and potential applications – the biggest mistake companies make is expecting AI to replace human judgment and involvement entirely.

Insufficient model testing and validation – AI systems require extensive testing and ongoing refinement. Organizations that deploy AI without adequate preparation are more likely to face issues around bias, output quality, and accuracy.

Not proactively correcting against false information – every AI system needs automated and human review processes to catch errors and ensure that output quality meets organizational standards. While much of the human involvement can occur during development and testing, you should also use commands to narrow program scope and parameters to reduce the risk of your tool being confused by contradictory information.

Implementing AI Tooling Across People Operations