AI in HR: stop flooding your stack, start focusing your strategy
AI is no longer optional for HR leaders. The risk is now two-sided: avoid the tech and fall behind, or chase every shiny tool and bury your team in noise.
Executive coach Jade Green and Anna Volkova, head of people and culture APJ at HiBob, are seeing that tension play out daily. Their message is blunt: focus beats FOMO. Human connection beats canned outputs.
The three camps - and the only one that wins
Green sees three groups emerging:
- Stuck: Paralysed by "don't know where to start."
- Over-indulgers: "Way too into it," implementing fast without a clear purpose.
- Strategic: A small, growing group tripling down on the right use cases. They win in 2026 and beyond.
The move now is a ruthless look at goals and work design. "What are our goals for 2026? What are the key things we need to do?" From there, run a task audit, then delegate the right portion of work to AI. Less tools, more intent.
Run a task audit before you buy your next tool
- List the work: Across each HR function (TA, L&D, ER, ops), map recurring tasks.
- Tag tasks: Automate, accelerate, or keep human. Be conservative for anything involving judgment, context, or emotion.
- Choose "a tool for the task": One clear job per tool. Avoid overlap.
- Redesign roles: Shift time from admin to "peopling" - the work only humans can do.
Green is clear: "I don't think more is better. I think less is more." The goal isn't headcount cuts. It's freeing people to do their best work, not busy work.
The hidden cost of too many tools
Some HR teams are "busy being busy" with AI - experimenting with multiple apps, copilots and platforms. It looks like innovation. It feels productive. It often adds friction.
- Set usage rules: Write simple guidelines for where, when and how each tool is used.
- Train live, in-flow: "In real time" with experts beats static courses that age out fast.
- Test hard before rollout: Pilot with a small group. Validate impact. Only then add to the core stack.
- Give every tool an owner: Scope, metrics, and a kill switch if value drops.
Don't outsource thinking - or humanity
Green is seeing leaders hand off their reasoning to GPTs and similar models, then struggle to defend decisions. That's an operational risk - and a cultural one.
Draw a line around human-only moments. "If you're doing performance management… pats on the back… do it authentically, do it real." Over-automate the emotional labor, and top performers will see the effort drop - and they'll leave.
- Human-only zones: Performance conversations, critical feedback, final decisions that impact pay or role, recognition.
- AI-assist zones: Drafting structures, prepping scenarios, data pulls, summarising notes - but the leader finishes the work.
If you've lagged, you're not late - but the window is closing
Staying out is no longer an option. Start narrow. Double down on your strengths.
- Define "success by Dec 2026": Pick clear outcomes for your function.
- Pick 1-3 use cases: Example: talent acquisition sourcing, internal mobility matching, policy Q&A.
- Choose 1-2 tools per use case: Own the setup, prompts, workflow and metrics.
- Ship a 30-60-90 plan: 30 = pilot, 60 = training and refinement, 90 = scale with guardrails.
Inside HiBob: expectation, experimentation and psychological safety
Volkova says HiBob leaned in early - in product and internally. There's an expectation that leaders and teams use AI, but it's treated as a managed experiment with clear parameters: how far it goes in daily work, where coaching starts, and where the human element sits.
One smart move: translating their leadership framework - "the Code" (Connect, Own, Disrupt, Evolve) - into live practice through AI. Using tools like ChatGPT, they built prompts, coaching scenarios and development plans tied to those habits and to company values. It gives leaders a starting point; HR then overlays nuance and context. Conversations move faster - and get better.
Culture work is deliberate: clarity, capability, community. AI usage is part of performance reviews in a developmental way. Teams get accessible tools, live training, rituals like AI Day, and departmental champions. They are also asking the hard questions for early-career talent: what experiential learning needs to be protected because AI can't provide it?
Your 2026 mandate
- Start with outcomes: Goals first, tools second.
- Adopt fewer tools: Be hyper specific about the job each one does.
- Guardrails: Usage rules, owners, testing, and metrics.
- Train in the flow of work: Prefer live, on-the-job enablement over static courses.
- Protect human moments: Performance, coaching and recognition stay personal.
- Measure and model: Include AI usage in performance. Leaders go first.
Helpful resources
- NIST AI Risk Management Framework - a practical structure for governing AI use.
- Complete AI Training: Courses by Job - curated learning paths to upskill HR teams without tool sprawl.
The edge isn't more AI. It's strategic AI - paired with leaders who still look people in the eye, do the hard conversations, and make clear calls. Keep the stack lean. Keep the connection real.
Your membership also unlocks:
