AI & Future of work - 2 May 2025

Welcome to FullStack HR, and an extra welcome to the 109 people who have signed up since last edition.

If you haven’t yet subscribed, join the 9200+ smart, curious and like-minded HR people by subscribing here:

---

Happy Friday!

Lift got in the way last week but this week, I’m back with a new edition that covers the two last weeks.

I hope you enjoy it!

Also, if you want me to swing-by your people team to do an update on all things AI & HR - reach out!

---

1. Microsoft 2025 Work Trend Index: “The Frontier Firm” Arrives
   Microsoft’s annual study of 31 000 workers, LinkedIn hiring data and Microsoft 365 telemetry says 82 percent of executives expect 2025 to be the year they rebuild jobs around autonomous AI agents. The report talks about “Agent Boss” and “Intelligence on Tap” and profiles AI-native start-ups already slashing meeting hours and doubling revenue per employee.
   
   Why it matters for HR: This is not an isolated tech trend, it is a hard deadline. HR will have to redesign job architectures to show where an agent takes over a task and where human judgement stays in play. Compensation models that reward hours or activity quickly lose meaning when agents handle bulk work. Skills frameworks must expand to include agent selection, prompt engineering and “oversight IQ”. Finally, every people process from requisition approval to exit interview needs a risk lens for agentic bias and security.
   Read more
   

2. IG Group Doubles Productivity With Anthropic’s Claude for Work
   A new case study shows the global financial-services firm getting a 100 percent productivity lift, reclaiming 70 analyst hours per week and hitting full ROI in under three months after rolling out Claude.
   
   Why it matters for HR: This is rare, public, bottom-line proof. HR can use the numbers to persuade finance chiefs that AI pilots are not science projects but EBITDA plays. The phased rollout—manager training first, usage targets second offers a ready-made change blueprint. HR business partners should adapt that cadence for other functions, set leading metrics (prompts per employee, agent-assisted cycle-time) and bake them into performance scorecards.
   Read more
   

3. GitHub Copilot Edits Reaches General Availability in JetBrains IDEs
   GitHub has introduced Copilot Edits GA, letting developers select multiple files and ask the AI to refactor, optimise or comment code instantly.
   
   Why it matters for HR: A week of refactoring is now one prompt. Engineering capacity models, promotion criteria and career pathways must shift from lines-of-code to code-review, architecture and AI-orchestration skills. Recruiters should start flagging “pair-programmer coach” experience. L&D needs micro-modules on safe AI code review, while reward teams revisit incentive plans to value design quality over raw throughput.
   Read more
   

4. Slack CLI 3.0.x and Docs Go Open Source
   Slack’s platform update moved the CLI and its documentation to public GitHub repos, fixed upgrade bugs and invited community pull requests.
   
   Why it matters for HR: Slack is many companies’ unofficial HR front door. Now any HR or COE specialist can script recruiting funnels, policy chatbots or onboarding journeys without waiting for IT. That democratises automation, so HR should launch a “citizen-builder” guild, issue coding guardrails and recognise contribution badges in performance reviews. The open-source docs also lower vendor lock-in risk, a plus for compliance.
   Read more
   

5. IBM: “Becoming an AI-First Enterprise Starts in HR”
   IBM CHRO Nickle LaMoreaux lays out how AskHR migrated to watsonx Orchestrate and now resolves 94 percent of queries autonomously while posting an NPS of +74.
   
   Why it matters for HR: The article names the uncomfortable moves: shutting down the HR help-line to force adoption, retraining staff as “agent governors”, and hard-coding ethical checkpoints. HR leaders can borrow the maturity model (self-service → assisted → predictive → autonomous) for their own roadmaps. It also signals that employee-experience scores can rise even when human touch is replaced by AI provided governance is tight and data flows are transparent.
   Read more

6. Paychex Survey Spurs Debate on AI-Driven Hiring Risks
   Rochester Business Journal’s 24 April report cites a Paychex survey showing 65 percent of small-business HR leaders already use AI in hiring and 53 percent plan more investment, but legal experts warn of algorithmic-bias exposure and evolving disclosure laws.
   
   Why it matters for HR: Mid-market firms are adopting AI faster than many enterprises, which means case law could be shaped in their courts first. HR must therefore own an “audit-ready” data trail, documented prompts, model versions and diversity outcomes before regulators ask. The findings also widen the talent gap: candidates will expect AI-fast processes everywhere. Ignoring that expectation risks brand damage and offer declines.
   Read more
   

7. KPMG Pulse Survey: AI-Agent Deployments Stall at Pilot Phase
   A KPMG research finds only 11 percent (!) of billion-dollar companies have scaled agents to production despite surging pilot activity.

   
   Why it matters for HR: The bottleneck is not tech but change management. HR can unblock scale by inserting agent adoption targets into OKRs, tying bonuses to safe deployment milestones and codifying “human-in-command” protocols. The survey’s stage model (experiment, pilot, deploy) doubles as a competency ladder for internal certifications, use it to craft promotion paths for AI power users.
   Read more

8. Google AI Works Pilot: 122 Hours Freed per Employee
   Google’s 24 April report on its UK “AI Works” pilot shows that simply allowing staff to use gen-AI for everyday admin – plus a two-hour upskilling session – let workers reclaim an average 122 hours a year. Adoption jumped most among groups usually left behind: weekly AI use by women over 55 rose from 17 percent to 56 percent in just three months. The study estimates UK productivity could rise by £400 billion if similar training rolled out nationwide.
   
   Why it matters for HR: The numbers give HR a concrete, time-savings benchmark to sell AI enablement to executives: four working weeks per employee back on the calendar. The pilot also proves that quick permissioning and bite-size training smash the “I’m not allowed to use AI” barrier that still holds many teams hostage. HR can replicate the formula by (1) writing a one-page “yes-you-may” AI policy, (2) running micro-clinics on prompt basics, and (3) tracking reclaimed hours as a hard ROI metric. The inclusion dividend is just as important – targeted coaching can close adoption gaps for older and lower-income workers, turning AI rollouts into both a productivity play and a DEI win.
   Read more
   

9. Josh Bersin: “The End of HR As We Know It?”
   In his 26 April essay, Bersin argues that AI signals the end of HR as we know it. I really value Josh and his insights, yet most of us working with AI and HR spotted this shift back in the fall. The arguments he presents align with many of the points I’ve been raising for the past year. Nonetheless, he remains an influential voice in the HR space, and I agree with his conclusions.
   
   Why it matters for HR: The piece shifts the burden of proof: HR must present system-level ROI models, not pilot anecdotes. That means building a data backbone, upskilling analysts in storytelling and adopting portfolio management for workforce initiatives. Use the essay to justify a holistic AI budget that spans recruiting, learning, analytics and culture, rather than scattered line-item requests.
   Read more
   

10. Forbes Tech Council: AI Agents Reinvent Business Analytics
    A column explains how autonomous agents compress analysis cycles from weeks to minutes and offers a three-step roadmap, audit data, define guardrails, deploy agents in “dual mode” with humans to get started.
    
    Why it matters for HR: When dashboards refresh themselves, the people-analytics team becomes a storytelling and change-management engine. HR should redeploy analysts toward behavioural science, nudging design and narrative creation while letting agents handle data wrangling. It also frees capacity for proactive interventions, spotting flight-risk cohorts or bias patterns before they hit the KPI dashboard.
    Read more