76% of SMBs Expect to Increase AI Use, But Few Feel Ready on Skills, Harvard Business Review Analytic Services-TriNet Survey Finds

SMBs Are Accelerating AI Adoption-HR Needs a Clear Skills Playbook Now

A new Harvard Business Review Analytic Services survey sponsored by TriNet shows a fast shift: 76% of SMBs expect to increase AI use in the next 12 months, yet only 19% feel highly prepared to recruit or develop the AI skills they need. That's a wide gap-and it lands squarely on HR's plate.

"AI is fundamentally transforming the way SMBs operate," said Catherine Wragg, Chief People Officer at TriNet. The takeaway for HR: move fast on skills, role design, and practical enablement-or teams will adopt AI unevenly and expose the business to risk.

Key findings HR should act on

• 56% expect AI will require them to develop or train employees differently.
• 49% anticipate changes in existing roles and responsibilities due to AI.
• 56% expect difficulty determining which AI skills their organization actually needs.
• 55% say one of the most in-demand AI skills will be hands-on experience using AI tools to accomplish work tasks.
• 70% report AI is increasing demand for human capabilities like creativity, intuition, and discernment.
• 49% anticipate difficulty upskilling existing employees on AI; 37% expect challenges evaluating candidates' AI skills.
• 79% agree AI is driving the need to upskill current talent.

What this means for HR

• Shift hiring and development from job titles to skills. Build a targeted skills taxonomy that blends AI tool fluency with human judgment.
• Update job architecture and role profiles to reflect human-AI collaboration (who does what, with which tools, and how quality is verified).
• Equip managers to coach AI use in day-to-day workflows, not just in theory.
• Stand up lightweight AI governance: usage guidelines, privacy rules, prompt hygiene, and review steps for sensitive outputs.
• Measure productivity and quality outcomes, not just "training hours."

A 90-day HR action plan

• Days 1-30: Identify 3-5 priority workflows per function (e.g., content drafting, reporting, customer responses). Define success metrics (time saved, error rate, quality score). Publish acceptable-use guidelines and data protections.
• Days 31-60: Pilot AI use with volunteer teams. Provide role-based playbooks (prompts, checkpoints, red flags). Launch a simple assessment to baseline AI skills and set learning paths.
• Days 61-90: Formalize skill rubrics for hiring and internal mobility. Roll out microlearning, office hours, and a champions network. Report results and expand pilots to adjacent teams.

Build the right skills mix

• AI tool fluency: prompt writing, document analysis, summarization, data cleanup, and basic automation.
• Human strengths: creativity, problem framing, domain expertise, stakeholder empathy, and ethical judgment.
• Team practices: version control of prompts, quality reviews, citing sources, and documenting decisions.

Hiring and assessment-what to change now

• Job posts: Ask for "experience using AI tools to deliver work outputs" with 1-2 concrete examples relevant to the role.
• Work sample tests: Give a realistic task, a small data or content set, and evaluate the approach: prompt design, accuracy, bias checks, and rationale.
• Portfolios over buzzwords: Request artifacts (before/after outputs, prompt libraries, QA checklists) instead of generic "AI experience."
• Structured interviews: Score candidates on judgment, exception handling, and how they verify AI outputs.

L&D that actually sticks

• Microlearning tied to real tasks (10-15 minutes), plus "apply it this week" challenges.
• Community of practice: peer demos, prompt swaps, brown bags. Reward useful shared templates.
• Manager enablement: coaching questions, weekly usage reviews, and a simple rubric for output quality.
• Track outcomes: time saved, error reductions, cycle time, customer scores-not just course completions.

Governance and risk without slowing teams

• Clear rules for sensitive data, PII, and confidential content. Default to approved tools with audit logs.
• Marked AI-generated outputs that require human review before external use.
• Ethics checks for fairness and IP. Use short pre-release checklists.
• Reference frameworks like the NIST AI Risk Management Framework for common-sense controls NIST AI RMF.

Metrics to report to leadership

• Adoption: percent of roles using AI for defined tasks; prompt library usage.
• Performance: time to complete standard tasks; error rates; quality review scores.
• Talent: internal mobility into AI-adjacent roles; hiring time-to-fill; skill assessment uplift post-training.
• Risk: policy exceptions, data incidents, and rework due to AI errors.

Context and next steps

The signal is clear: SMBs plan to expand AI use fast, but capability building is lagging. HR can close the gap by making skills visible, enabling day-to-day workflows, and protecting the business with light, practical guardrails.

For additional context on the research, see Harvard Business Review Analytic Services.

Need practical resources for HR-led upskilling and implementation? Explore AI Learning Path for HR Managers and AI for Human Resources.

About the study

The findings above are based on "The New Talent Playbook for Small and Midsize Businesses in the Age of AI," a survey of 230 respondents familiar with their SMB's U.S. talent practices, conducted by Harvard Business Review Analytic Services and sponsored by TriNet (NYSE: TNET).