Why 81% of Leaders Are Deploying AI Agents in 2025

Presented by

AGENTIC HR: WHY 81% OF LEADERS ARE DEPLOYING AI AGENTS IN 2025

AI agents are no longer futuristic. They’re now automating tasks, recommending hires, drafting policies, and even managing internal talent systems.

According to the 2025 Work Trend Index Annual Report

• 82% of leaders expect to use agents to meet workforce capacity needs in the next 12 months

• A workforce of “hybrid” teams of humans and agents will enable companies to be more agile, scale more efficiently, and generate value faster.

• However, only 12% of leaders have moved past pilots.

You may already be behind if your HR team still treats GenAI as a side project. 

In 2025, your most strategic coworker might not be human.

Inside this edition

• What Agentic AI means for HR and why it’s gaining traction

• 5 practical steps to scale AI agents responsibly

• How Wells Fargo, Dow, and Bayer approach agent deployment

• Legal and governance risks to address early

• Bonus: Free blueprint to launch your first AI agent team

📥 Free Resource: Download your AI Agent Deployment Blueprint to get sample use cases, governance tips, and tech stack templates.

🗞️ In case you missed it: From a rollback on nonbinary reporting to rising stress, revived tariffs, and a White House workforce overhaul, this week’s headlines show how policy, tech, and power are redrawing work boundaries in real time.

Read on to discover what Agentic AI can—and can’t—do for your team (yet)...

🍀Workplace Well-being and Mental Health Summit

🌟 Save your seat for our May event: The Workplace Well-Being & Mental Health Summit! The event is free, but with a premium Hacking HR membership ($159 per year), you can earn credits for this (11.25 SHRM + 11.25 HRCI) and all our upcoming events, plus access to all the PCSI certificate programs and crash courses. Learn more here.

🍀Referral program

💸 Help Us Grow and Earn While You Do

The Hacking HR Affiliate Program offers a great opportunity to earn a 20% commission on Premium Hacking HR memberships.

1. Simply access the Hacking HR Lab via hackinghrlab.io (sign up for a free or Premium plan, or log in if already signed up).

2. Navigate to the Affiliate Program dashboard in the main menu.

3. Get your Affiliate link and start creating content to promote Hacking HR!

4. Where to share: LinkedIn posts, articles, groups, personal blog, discussions in industry forums, and on a subtle placement in your email signature.

SPREAD FOR YOU
📢 MEET THE NEW DIGITAL CO-WORKERS

Most HR teams are familiar with Generative AI tools, such as chatbots that answer questions, summarize content, or draft emails on command. All these tools need direct human prompts to work. They're helpful, but limited.

AI-powered workflows move a step further. After a trigger happens, they automate sequences of tasks (like sending onboarding emails, scheduling meetings, or updating databases). But they still rely on rigid, predefined rules built by people.

AI agents are different. 

They think, plan, and act independently within a defined framework.

Agents don’t just respond to prompts or follow simple workflows; they analyze situations, choose actions, connect with external tools, and adapt over time based on results. They behave more like digital teammates than tools.

Here’s a simple way to visualize it:

Put simply:

• GenAI is a responsive assistant.

• Workflows are scripted choreographers.

• Agents are autonomous coworkers.
  

In HR, this means an agent could:

• Run a new hire’s onboarding from start to finish

• Adjust the onboarding steps based on feedback

• Suggest a buddy match from internal networks

• Update HRIS records automatically

• Recommend the next career path—all without asking for a prompt.
  

And that’s why HR must rethink how teams, tasks, and systems are designed. 

Yes, agents will boost productivity. But above all, they will imply an unprecedented workforce shift.

Next, let’s dive into how to deploy agents responsibly—and what pitfalls to avoid.

HR FRAMEWORKS

5 PRACTICAL STEPS TO SCALE AI AGENTS RESPONSIBLY

Deploying AI agents is far more complex than launching traditional SaaS tools.

Unlike apps that automate single workflows, agents operate across multiple organizational layers (data sources, workflows, decision frameworks) and external systems, autonomously making decisions, adapting to new data, and acting without constant human input​.

As Edwin van Bommel (PwC) warns, "the more autonomously these digital assistants operate, the more important it becomes to maintain control over their choices and underlying principles".

Ethics, transparency, and reliability must be baked into every deployment decision.

Here’s a step-by-step plan HR leaders can use to roll out AI agents while mitigating risks:

1. Build Your Agent Council

An “Agent Council” is a cross-functional team that sets guardrails before design work starts. In Microsoft’s own AI adoption practices, cross-functional councils were essential in establishing governance frameworks before Copilot agent deployments.

The company's CEO can establish this team to align business, technology, legal, and compliance requirements and ensure that agent deployments fit strategic priorities and legal frameworks.

• Champion the creation of the Agent Council and lead coordination.

• Define workforce-related risks and ethical standards.

• Secure representation from Employee Relations and DEI teams.

2. Define Agent Roles

Agents must be tied to business value, not just technical possibilities. A great way to start is by spotting where AI agents can dramatically reduce friction points. Wells Fargo, for example, deployed an agent supporting 35,000 bankers by clearly defining its role: to retrieve operational information in seconds.

HR must work with the Council to define agent "job descriptions" based on critical workflows.

• Identify pain points in workforce workflows that agents could optimize.

• Write outcome-focused agent role profiles (e.g., onboarding specialist).

• Set up success KPIs for each agent role (time saved, accuracy improved).

3. Control Agent Access and Learning Scope

Early Microsoft pilots restricted Copilot agents to structured content only before granting broader access to sensitive HR data. Similarly, when an organization deploys early agents, it must ensure they operate in a "sandbox" environment with limited access to reduce risk.

• Approve agent access only to HRIS fields and documents strictly necessary for their tasks.

• Collaborate with IT to monitor interactions and system activity.

• Define thresholds for agent learning (e.g., what external data sources agents may analyze).

4. Appoint Dedicated Agent Managers

Every agent needs human supervision. HR must institutionalize the "Agent Manager" role to monitor, coach, and correct agent actions. For example, at Wells Fargo, managers were appointed to oversee agent outputs, improving quality control and escalations across 4,000 branches. It is recommended that HR leaders take the same approach.

• Define the Agent Manager job profile and selection criteria.

• Train managers in agent monitoring and ethical escalation procedures.

• Integrate agent performance into team KPIs.

5. Embed Governance, Ethics, and Monitoring from Day One

Governance is not a Phase Two project. Companies should include live tracking of agent actions, explainability dashboards, and human checkpoints from day one. HR must lead ethical oversight and design transparency requirements into the agent system from the beginning.

• Co-develop dashboards to track agent activities and decision justifications.

• Mandate audit logs for every critical agent interaction.

• Set up a regular cadence of compliance and ethics reviews.

⚠️ Note: Legal and Governance Risks to Address Early

Embedding agents into HR systems without a strong legal and governance framework can expose organizations to serious risks, including privacy violations and biased decision-making. These deployments must meet high compliance and ethical standards, and there are four critical risks HR leaders must address before scaling AI agents:

• Bias Amplification: Agents trained on historical data can unintentionally reinforce existing inequities in hiring, promotions, or performance evaluations. Regular audits and bias testing are essential.

• Data Privacy Breaches: Agents that access employee records, health information, or salary data must operate within strict privacy parameters to comply with GDPR, CCPA, and local labor laws.

• Accountability Gaps: Without clear human oversight, agents making decisions about people processes can trigger legal liabilities. Every agent must have a designated human "manager" responsible for outcomes.

• Explainability Failures: Agents must be able to explain why they made a decision. Inability to provide decision rationales could breach emerging AI transparency regulations and erode employee trust.

Building strong governance is the foundation for responsible innovation. Set guardrails early to empower your agents without losing control.

Learn more:

AI Agents: Your New Digital Employees — PwC Report (2025) — Ethics, autonomy, and agent deployment best practices.

How to Deploy AI at Scale — PwC and Microsoft (2024) — Governance and accountability models.

Agentic AI: The New Frontier — Deloitte (2025) — Strategies for scaling AI agents responsibly.

Most teams are experimenting. Few are scaling. Where would you place yours?

POLL

✨ Want to go deeper? Join us at the AI Experience Summit for strategy, tools, and real-world stories.

UPCOMING EVENTS
THE AI EXPERIENCE SUMMIT

The three-day AI Experience Summit is a must-attend for HR, tech, and business leaders building the next chapter of work.

• 📅 When: June 3-5, 2025 (7 am–11 am Pacific Time)

• 💻 Event Type: Online | Free

• 📢 Content Format: Panel | Workshop

Continue reading and learn lessons from Wells Fargo, Dow, and Bayer.

SUCCESS STORIES
WHAT HAPPENS WHEN YOU DEPLOY AGENTS THE RIGHT WAY

Did you know the modern employee faces an average of 275 interruptions daily? Disruptions include meetings, chats, emails, and pings.

Scaling AI agents offers a way to reduce friction, automate repetitive tasks, and free up human focus for more meaningful work.

The following organizations show how strategic agent deployments are already driving measurable results.

🏦 Wells Fargo: Scaling Human-Agent Collaboration in Banking

Impact:

📊 245 million interactions handled in 2024 alone, doubling original projections​

🔒 0 sensitive customer data exposed to external LLMs due to Wells Fargo’s custom privacy-first orchestration layer​

🧠 Building multi-agent systems for tasks like re-underwriting 15 years of loan archives autonomously​

For Wells Fargo, handling millions of customer interactions while ensuring compliance with strict privacy regulations posed a major challenge. Existing customer support models couldn't scale efficiently without risking data exposure, so they invested in Agentic AI:

• Deployed Fargo, a virtual assistant powered by Google's LLMs, operating through a custom-built Tachyon AI platform that scrubs and tokenizes user inputs before invoking models.

• Designed a privacy-first orchestration pipeline: no personal identifiable information (PII) ever reaches the LLM.

• Expanded to multi-agent architectures internally — for example, re-underwriting loan archives through coordinated agent networks, drastically reducing manual workloads.

⚡ Dow: Millions Saved by Scaling Freight Agents

Impact:

💸 Millions of dollars in projected first-year freight savings​

📄 100,000+ PDF invoices annually processed autonomously​

🧠 Over 100 additional agent use cases identified across supply chain operations​

Dow’s global shipping operation generated thousands of invoices daily, each with potential billing errors hidden in massive unstructured datasets. Manual auditing was too slow, costly, and inconsistent. Here’s what they did to overcome these challenges using AI Agents:

• Built an autonomous agent to extract, structure, and audit invoice data automatically.

• Launched a Freight Agent allowing employees to query billing anomalies in natural language.

• Achieved early success by uncovering hidden cost discrepancies within minutes instead of weeks.

🌱 Bayer: Transforming Internal Communications with AI Agents

Impact:

🌎 19,000 unique users across nine languages within two months​

📈 Over 80% employee satisfaction rate with the AI agent​

🗣️ AI agent integrated into daily operations and team meetings​

Bayer launched a major organizational transformation based on a new operating model called Dynamic Shared Ownership (DSO).

Communicating this change across a global, multilingual workforce, with critical documents scattered across different formats and systems, posed a significant barrier to adoption and clarity. How did they use AI agents?

• Developed DSO Go, an AI Agent built on Cognigy.AI + Generative AI, capable of answering dynamic employee queries across multiple formats (PDFs, Word docs, PowerPoints).

• Deployed the agent in nine languages, ensuring consistent communication across Bayer’s global footprint.

• Embedded the agent into daily meetings and team workflows, making it an integral tool for goal-setting and operational alignment.

Inspired? Grab the resource and let’s get to work!

YOUR DOWNLOADABLE RESOURCE
AGENT DEPLOYMENT BLUEPRINT FOR HR

This guide helps HR leaders and teams go from AI theory to action—safely and strategically. Based on insights from Microsoft, Deloitte, and PwC, the blueprint includes:

✅ A readiness checklist to assess where you stand

✅ A governance framework to define oversight and accountability

✅ A step-by-step launch flow to pilot your first agent

✅ Use cases by HR function: recruiting, onboarding, L&D, and more

✅ BONUS: The Agent Boss Self-Assessment

Before we wrap up, here’s a quick look at the latest headlines.

TOP PICKS FOR YOU
🗞️ IN CASE YOU MISSED IT

Here’s what’s making waves right now.

📝 EEOC Nixes Nonbinary Option in 2024 EEO-1 Filing, Drawing Backlash

The US Equal Employment Opportunity Commission will open its 2024 EEO-1 Component 1 data collection portal from May 20 to June 24. Still, in a controversial shift, it has removed the option to voluntarily report nonbinary employees, previously allowed in a comment box. Read more →

🚨 Google Draws the Line: Return to Office or Risk Losing Your Job

In a major shift, Google is mandating that remote employees return to the office at least three days a week or face voluntary exit offers and potential job cuts, as the company focuses on cutting costs, boosting in-person collaboration, and investing heavily in artificial intelligence amid broader tech industry moves away from remote work. Read more →

🩺 Trump Administration Boosts Medicare Advantage Payments but Insurers Still Voice Concerns

The Trump administration has raised Medicare Advantage payment rates by 5.06% for 2026 — higher than in previous years — but insurers like UnitedHealth, Elevance Health, and Centene say it only partly offsets past underfunding, while CMS also introduced new rules streamlining prior authorization and rebranding its health equity efforts. Read more →

🔧 White House Orders Overhaul of Workforce Programs to Boost Skilled Trades

President Trump signed an executive order to revamp federal workforce development by consolidating programs, expanding apprenticeships, promoting alternative credentials, and preparing American workers for high-paying, skilled trade jobs driven by reindustrialization and emerging technologies. Read more →

🛑 Sellers Hit Pause: Trump Tariffs Dampen Amazon Prime Day Hype

Amid soaring tariffs on Chinese imports imposed by former President Trump, many Amazon third-party sellers are scaling back or entirely skipping Prime Day 2025 to protect their razor-thin profit margin. Read more →

🛑 White House Blasts Amazon Over Planned Tariff Transparency Move

The White House labeled Amazon "hostile and political" after a report suggested the e-commerce giant was considering listing tariff costs separately for shoppers amid Trump’s new trade policies. However, Amazon later clarified the plan was rejected internally — a flashpoint in the complex, evolving relationship between Trump, Amazon, and Jeff Bezos. Read more →

⬇️ NEW Gallup Report Shows Global Employee Engagement Drops, Stress Rises

Gallup's 2025 State of the Global Workplace report reveals a troubling two-point drop in global employee engagement to 21%, the lowest since 2021, costing the global economy an estimated $438 billion in productivity, as declining manager morale, remote work stress, and widening disconnects between leadership and employees undermine workforce wellbeing. Read more →

Sponsor Daily Jam

Would you like to be featured in Daily Jam? Place your brand alongside deep industry research, actionable tips, and useful resources for HR and other leaders.