How AI Is Re-ranking Soft Skills in the Workplace

We have been doing research at Draup on soft skills in the AI-shaped workforce, and I wanted to share an early look at it. One note up front: this is a working paper, with early results from a model we are still refining. Please read it as a preview of where the evidence points rather than a final benchmark.

The short version is this. As AI takes on more of the drafting, analysis and routine execution in a role, the human skills that direct, judge and translate that work decide whether a transformation delivers value or simply produces more output. The brief finds that AI is re-ranking soft skills in the workplace, and that broad labels such as communication or critical thinking are too coarse to show how.

To see this clearly, we built the Draup Soft Skills Model: 183 soft skills in 13 domains, drawn from AI transformation trends and mapped to the 3,553 roles in Draup's Soft Skills by Role library. The brief walks through these patterns using two example roles, a Marketing Manager and an HRBP Manager.

Why broad soft skill labels are too coarse to show what AI is changing

Most soft-skill frameworks stop at a few dozen broad labels such as communication, leadership, critical thinking and adaptability. The brief argues that those labels are too coarse to show what AI is changing. Two roles can both need communication while one needs Executive Communication and the other needs Intent Articulation, the ability to brief an AI tool on what good looks like. A competency model that cannot tell them apart cannot tell a hiring manager what to screen for or an L&D team what to build.

The model starts from the trends in AI transformation, meaning what models now draft, what they feed and where people still decide, and asks which human skills those shifts make high-impact. Each of the 183 skills is defined precisely enough to act on and is assigned to at least one role. Draup then mapped the library to every role in its Soft Skills by Role library, so each of the 3,553 roles carries its own profile drawn from the same vocabulary.

The brief offers illustrative pairings between broad labels and the specific skills roles single out instead. In place of Critical Thinking, the model names Judgment, Professional Skepticism and Discernment. In place of Emotional Intelligence, it names Expectation Management, Trust Building and Conflict De-escalation. In place of Written Communication, it names Simplifying Complex Information, Data Storytelling and Intent Articulation. The brief notes that these pairings are illustrative rather than a formal hierarchy. Its practical point is that critical thinking cannot be screened for or taught as a single thing, while Professional Skepticism toward a confident AI output can.

Which soft skills are cited most across the 3,553 role profiles

In the brief, the 13 domains are ordered by how often their skills are singled out across the 3,553 role profiles. Digital & AI Readiness is one of the smallest domains, with six skills, yet it carries the two most widely cited skills in the library. Its most-cited skills are AI Output Evaluation, AI Delegation & Oversight and AI Ethics Awareness.

The brief groups the most widely cited skills into four clusters:

• Supervising AI, led by AI Output Evaluation and AI Delegation & Oversight.
• Translating analysis into action, led by Simplifying Complex Information, Data Storytelling and Executive Communication.
• Vetting and judging, led by Judgment, Technical Credibility and Professional Skepticism.
• Reliable execution, led by Process Adherence, Safety Consciousness and Quality Orientation.

Across the library, knowledge roles now carry AI Delegation & Oversight and AI Output Evaluation as emerging skills, while physical roles largely carry none. The brief describes this as a supervisory core forming. The library also holds skills that few roles single out today, such as Automation Bias Awareness and Artificial Intelligence Literacy. Because the library is drawn from transformation trends and not only from today's job descriptions, it can name a skill before the market asks for it, and the regular refresh shows when that skill starts to register.

How each role profile separates differentiating, emerging and universal skills

Each role in the library sorts its soft skills into three core tiers. Differentiating skills set strong performers in the role apart. Emerging skills are needed because AI is changing the work. Universal must-haves are expected in most jobs. A skill is marked emerging when it appears in a rising share of the role's current job descriptions and is tied to a specific AI-driven change in the work, and every assignment is checked against active job descriptions for the role.

The Marketing Manager and the HRBP Manager were selected from the full run because they most clearly illustrate the library-wide patterns for knowledge roles. The brief is explicit that they illustrate the findings and are not the evidence for them. Both roles carry four emerging skills. For the Marketing Manager, each emerging-skill rationale is about briefing, choosing among, overseeing or reviewing AI-generated work. For the HRBP Manager, the emerging tier adds ethics and data-handling skills alongside AI oversight and evaluation.

Translation skills now set roles apart outside communications

Data Storytelling is a differentiator for both the HRBP and the Marketing Manager, and Simplifying Complex Information is a differentiator for the HRBP. Each rationale has the same shape: a technical input such as people data, marketing analytics or HR frameworks, and a non-specialist audience who must act, such as business leaders, senior leadership or line managers.

The brief's reading is that when AI produces the analysis, the human's scarce contribution is making it land with someone who has to decide. On that reading, translation moves from a pleasant extra to the thing that justifies the seat.

How the human job is shifting from making to judging

The emerging-skill rationales share a vocabulary: verify, check, monitor, correct, review and oversee. The HRBP must check AI-generated workforce analytics against business context, and the Marketing Manager must review AI content for brand accuracy and tone. Both rationales describe judging something made rather than making something.

The brief argues that this raises the stakes on established skills. In the HRBP profile, Holding Others Accountable and Courageous Conversations are described in terms of team and employee-relations work. Read alongside the emerging tier, they are also the skills a manager needs to challenge a fast, confident and sometimes wrong contributor, human or AI. The HRBP's AI Ethics Awareness rationale names the risk as treating AI output as neutral. In the brief's framing, accountability is transferring without authorship: the model drafts, and the person remains answerable.

What the brief recommends for hiring, L&D and workforce planning

The brief is written for CHROs and the talent, L&D and workforce planning leaders who work with them, and it makes five recommendations:

• Common language. Use the 183-skill library as the shared vocabulary across hiring, L&D and planning, so that a gap found in workforce planning maps directly to an interview question and a learning intervention.
• Hiring. Screen knowledge roles for the delegation-and-evaluation loop directly by asking how the candidate decides what to hand to a tool and how they check what comes back. Add intent articulation and creative judgment for creative roles, and ethics awareness and data handling for people- and data-sensitive ones.
• Learning and development. The emerging tier is behavioral, so it is taught through practice on real tasks rather than technical courses. The brief names output evaluation as the highest-leverage single intervention, because every AI-exposed role needs it and every rationale ties it to a concrete risk.
• Job architecture. Keep the tiers separate. Rewriting a differentiator as emerging overstates disruption, and leaving the emerging tier empty for a knowledge role understates it.
• Workforce planning. Across the library, the number of emerging skills a role carries is a usable proxy for AI exposure. Knowledge roles where AI drafts or feeds the work carry several, while physical roles largely carry none, so reskilling budgets and productivity assumptions should differ across those groups and not just across functions.

Two limits apply to all of this. The results come from a model that is still being refined, so individual tier assignments may shift as it matures. The emerging tier is also, by definition, the part most likely to change: a skill marked emerging today may read as established within a few years, which is why the library is refreshed regularly.