From Adoption of AI to Agentic AI
I reviewed the HBS Working Knowledge – “AI in 2026: From Adoption to Agentic” paper. It brings together five Harvard Business School research perspectives on how AI is changing leadership, jobs, teamwork, decision-making, and talent strategy.
(Source: HBS-Working-Knowledge_AI-in-2026.pdf)
Executive summary
1. From GenAI to Agentic AI
The most important shift is from AI as a tool that responds to prompts toward AI as an agent that can plan, reason and execute workflows semi-autonomously.
For leaders, HBS describes a potential “digital support team” consisting of AI agents acting as:
Competitive intelligence analyst – continuously monitors competitors and external signals.
Chief of staff – analyzes calendars, meetings and priorities to identify where leadership attention is actually going.
Executive coach – anticipates questions, objections and feedback from stakeholders.
Strategic research assistant – continuously synthesizes information and translates it into potential actions.
The critical point is that agentic AI is not “set it and forget it.” HBS recommends a Plan → Execute → Learn cycle with human oversight at critical decision points.
2. AI is more acceptable as an augmenter than a replacer
The research on AI and employment produces an interesting distinction.
People are broadly comfortable with AI helping humans:
94% support using current AI to augment human work.
This increases to almost 96% for a more advanced future version of AI.
However, respondents were much more selective about AI replacing humans. They supported automation for roughly 30% of occupations based on current capabilities, rising to 58% when imagining substantially more capable AI.
This suggests an important management principle:
The question isn’t simply “Can AI do the job?” but “Should AI do the job?”
Human presence can remain strategically important because of trust, customer expectations, ethics and organizational legitimacy.
(HBS-Working-Knowledge_AI-in-2026.pdf)
3. The strongest model may be Human + AI + Human Team
One of the most compelling findings comes from the P&G field experiment involving 791 professionals.
AI-enabled teams generated significantly better ideas:
Top-10% ideas were 3× more likely to come from AI-enabled teams than individuals working without AI.
An individual with AI could perform at approximately the level of a two-person human team without AI.
AI helped less-experienced employees achieve performance closer to experienced colleagues.
AI-enabled workers reported more enthusiasm and energy and less anxiety/frustration.
The implication is particularly relevant for transformation and project management: AI doesn’t simply increase individual productivity—it can change the structure of collaboration and break down functional silos.
HBS therefore recommends treating AI as a teammate rather than merely a tool, while using AI-enabled individuals for efficiency and AI-augmented teams for breakthrough innovation.
(HBS-Working-Knowledge_AI-in-2026.pdf)
4. In high-stakes decisions, performance builds trust
The banking research is especially interesting.
In a study involving 9,000 participants, respondents were:
4.3 percentage points more likely to select a human bank manager rather than an algorithm for loan decisions.
7.6 percentage points more likely to select a human for pretrial-release decisions.
But the deeper finding was that decision quality and effectiveness mattered more than fairness in the experimental choices. Approximately one-third of respondents actually preferred algorithms, considering them more fair or effective than humans.
The strategic implication for financial institutions is powerful:
AI adoption in decision-making will depend not only on technological capability but on demonstrable performance, transparency and trust.
(HBS-Working-Knowledge_AI-in-2026.pdf)
5. AI makes talent density more important
Boris Groysberg’s perspective moves the discussion from AI technology to organizational design.
His central argument is that hiring exceptional AI talent is not enough.
Companies need to:
Build talent density, particularly in critical roles.
Design teams intentionally.
Integrate newly acquired talent into the organization.
Recognize that successful people at one company may not perform identically somewhere else because culture, processes and ways of working differ.
Use AI to identify skills gaps and accelerate reskilling.
Recognize that AI can amplify both excellent work and poor work.
Perhaps the most important conclusion is that organizations are likely to become smaller and more productive, rather than simply replacing every employee one-for-one with AI.
And importantly, soft skills and experience remain critical. AI can provide information and accelerate learning, but leadership still depends on judgment, influence, relationships and accumulated experience.
(HBS-Working-Knowledge_AI-in-2026.pdf)
From | To |
AI as a tool | AI as an agent / teammate |
Individual productivity | Human-AI collaboration |
Job replacement debate | Human augmentation + selective automation |
AI capability | AI + trust + governance |
More technology | Smaller, denser, higher-performing organizations |
The paper’s strongest message, in my view, is:
AI will not simply change what people do. It will change how organizations are designed.
That is particularly relevant to Project Management, Transformation, PMO and executive leadership: the competitive advantage may increasingly come from designing the right human-AI operating model, rather than simply deploying more AI tools.
My takeaway:
The winners of the AI transformation will not necessarily be the companies with the most AI.
They will be the organizations that learn how to combine:
AI capability + human judgment + high-performing teams + effective governance.
For leaders in Digital Transformation, Project/Program Management and PMO, this creates a fundamentally different challenge.
We are moving from managing projects with AI tools toward designing organizations and workflows where humans and AI operate as an integrated system.
Comments