The CHRO's Guide to AI and Agentic AI

When software starts doing defined work, the questions that follow are human ones: what happens to the roles, how people are told, what skills the company needs now, and whether employees trust the process. These questions have an owner, and it is the CHRO. This guide sets out how HR leaders plan, redesign, reskill, and communicate through an agentic AI transition, and how to use AI in HR's own processes without losing trust.

Why is agentic AI a CHRO issue?

An AI agent that processes invoices, answers support tickets, or drafts contracts is doing work that was in someone's job description. FISTA calls these agents digital FTEs: units of capacity that do defined work at a measurable cost. That framing is useful precisely because it puts agents into the workforce conversation where the CHRO can plan for them, rather than leaving them as a technology project that surprises the organization later.

The CHRO's leverage is in four places: workforce planning, job design, capability building, and trust.

How does workforce planning change?

Capacity now comes from three sources, and a credible plan models all three.

Source	How capacity is added	Cost behavior	Where it fits best
Hiring	Recruit, onboard, develop	Fixed, rises with tenure	Judgment, relationships, novel problems, oversight
Outsourcing	Contract for volume or skills	Variable by contract	Surge capacity, specialized skills, non-core work
Digital FTEs	Build or provision agents for defined work	Variable per task, falling with scale	High-volume, rule-bounded, measurable processes

The planning question is no longer "how many people do we need" but "which work belongs to which source, and what human roles remain around the agents." FISTA's digital FTE workforce planning whitepaper works through the method.

How should jobs be redesigned?

Redesign before deployment, not after. For each affected role:

1. Separate defined work from judgment work. The defined path is what a procedure describes; judgment covers exceptions, relationships, and decisions the procedure does not anticipate.
2. Assign the defined path to the agent with an explicit handoff: what the agent passes to the person when it stops, and in what format.
3. Rebuild the human role around exceptions, quality review, supervising agent output, and the customer or stakeholder relationship.
4. Update expectations and measures. A role that supervises an agent handling thousands of cases is measured on exception quality and throughput, not on cases processed personally.
5. Involve the people who do the work. They know the exceptions, and their participation is what makes the redesign accurate and accepted.

Roles that keep the judgment become more valuable, because the volume they oversee is larger. The how to redesign jobs around AI agents guide covers the mechanics.

What capabilities does the workforce need?

Most employees do not need to become engineers. They need to become good supervisors of AI output and good specifiers of work. The skills that matter:

• Process literacy: understanding how work flows well enough to describe it, because specification is what agents run on.
• Review and correction: the ability to check AI output against a standard and fix it, quickly and consistently.
• Judgment under ambiguity: handling the exceptions that agents escalate.
• Evidence habits: reading the metrics on an agent's performance and acting on them.
• Specification writing for team leads and analysts who define what agents should do.

Reskilling works when it is attached to a specific process change and a specific new responsibility. A general AI course catalog produces certificates; a program tied to "you will supervise the claims intake agent starting in March" produces capability. See how to reskill your workforce for agentic AI.

How should HR communicate the change?

Candor builds trust; vagueness destroys it. The communication should say, specifically:

• Which processes are changing, and when.
• What the agents will do, and what people will do instead.
• How roles, titles, and performance measures change.
• What happens to freed capacity: reinvested, redeployed, or reduced, and by what process.
• How employees can ask questions, raise concerns, and report problems with the agents.

If the answer to the capacity question is not yet decided, say that, and say when it will be. Employees can handle uncertainty; they cannot handle being managed around. The how to communicate AI changes to employees guide provides a structure.

How should HR use AI in its own processes?

HR operations are a strong candidate for agents: onboarding, benefits questions, policy answers, scheduling, document generation, and case routing all have volume and rules. FISTA's digital FTE for HR operations guide describes these uses.

HR decisions are different. Hiring, promotion, performance, and pay decisions carry legal exposure and fairness risk, and several jurisdictions regulate automated decision-making in employment. The safe pattern is assistance, not decision: a human remains accountable, AI output is one input among several, disparate-impact testing is done and documented, and candidates and employees are told when AI is used. This is general guidance, not legal advice.

What should the CHRO report?

Quarterly: roles redesigned and roles pending; digital FTEs in production and the work they absorb; capacity freed and its disposition; reskilling programs tied to process changes and their completion; employee sentiment on AI from surveys and question channels; and the controls applied to any AI used in HR decisions.

How can FISTA Solutions help a CHRO?

FISTA Solutions works with HR and executive teams through its AI enablement practice on workforce planning that includes digital FTEs, job redesign, and reskilling tied to real process changes, and builds the AI agents for HR operations that absorb defined work with human oversight designed in. Since 2017, FISTA has delivered 150+ projects for 50+ companies across 12+ countries.

To plan the human side of your AI program before the agents arrive, talk to FISTA on WhatsApp, or read the AI change management guide first.