Agentic Team: A New Way of Working Where Humans and AI Sit at the Same Table
I’ve long believed in the power of people and teams. People are motivated to create value; teams are accountable and hardworking. Yet what often slows teams down is the systems they operate within. In most organizations, the real problem isn’t capability or technology. Still, work eventually gets stuck and turns sluggish. Decisions are hard to make, feedback arrives late, and cross-team handoffs can take weeks. Individuals and teams try to move forward without always understanding the “why” behind their work. Agile transformations addressed many of these challenges—some organizations saw major improvements, others partial progress. The good news is that we are now entering a powerful new era that can help us unlock these and similar bottlenecks.
In this article, I want to describe the concept of agentic not as an organizational chart or another buzzword, but as a new team practice where humans and AI share responsibility.
What does “agentic” mean—at least from my perspective?
When people hear agentic, they often think of “autonomous AI.” I find that definition incomplete. I interpret agentic as a structure that can recognize a problem, take ownership, generate solutions, and turn those solutions into repeatable systems—not just one-off actions. Sometimes this structure is a person, sometimes a team, and sometimes a human–AI hybrid.
Agentic Teams:
- Own a clear problem and operate with a defined mission
- Have the authority to act without constantly looking upward for decisions
- Focus not on functions, but on customer problems and value
- Position AI not as a “support tool” but as an active team member, working within a multidisciplinary, agile structure
In other words, AI is not just a PowerPoint accelerator or an email writer. It is embedded in the workflow, part of decision-making, and integrated into feedback loops.
How does a human–AI hybrid team work?
Industrial-era organizations were built on a model where:
- Strategy was developed at the top
- Work was distributed downward
- Managers coordinated execution
This model made sense in a world where coordination costs were high and change was slow. Today, however, we see a different reality:
- Coordination costs are decreasing significantly
- Cognitive work like analysis, prediction, and simulation can be handled by AI
- The ability to create is becoming democratized (for example, almost anyone can now produce software—quality may vary, but access is widespread)
- The real bottleneck is the ability to build context and make decisions quickly
Across many industries, we’ve repeatedly observed increased productivity when agile team structures are adopted. Even in sectors where transformation is more challenging—such as manufacturing—bringing together people from production, quality, R&D, process, and operations around a clear problem and giving them decision-making space has shortened sample turnaround times, improved quality, and reduced costs. The shared sense of “we” that emerges from agile ways of working creates meaningful acceleration.
Now, to imagine what’s possible when AI joins the work, let’s extend this example. Picture a team developing chemical products. Assume the team includes people from R&D, marketing, product management, sales, supply chain, and production (at a representative level). In this model, AI doesn’t exist as a separate “role,” but as a component that supports specific workflows alongside the team:
- Formulation discovery and variation workflow: Scans past experiments, recipes, and performance results; generates and compares alternatives.
- Customer insight and feedback workflow: Analyzes field data, customer requests, and usage scenarios; feeds the team with meaningful market signals.
- Process scaling and manufacturability workflow: Simulates lab results without occupying production lines; highlights scalability, waste, and stability risks early.
- Quality and anomaly detection workflow: Continuously monitors sample, batch, and test data; flags deviations early and triggers proactive quality actions.
- Cost and margin simulation workflow: Calculates the impact of formulation and process changes on raw materials, energy use, and unit costs; makes commercial implications of technical decisions transparent.
- Regulatory and compliance workflow: Scans industry standards, customer certifications, and regulations; surfaces potential compliance barriers early.
Such a team can run the entire process internally—from the first spark of a need to sampling, from customer feedback to iteration. Working through shared human–AI workflows, the team can generate innovations far faster than before:
- AI scans past experiments and proposes ideas
- Humans decide, “Which option makes the most sense for us?”
- AI tests the outcomes
- The team selects the next experiment
- A sample goes to the customer
- The product improves based on feedback
- AI listens to the market
- The product moves toward commercialization
- Optimal cost–quality balance is maintained
- AI identifies alternative paths based on market feedback
What I’m trying to convey with this simple example is how the intelligence and agentic productivity that AI enables truly excites me. At the same time, this is new territory for all of us. I continue experimenting and learning in my own work. The ideas I’ve described will undoubtedly evolve and become clearer over time. Still, it seems likely that the organizations that stand out in the coming years will be those that can distribute authority without creating new bottlenecks, keep human judgment at the center while using AI, and build fluid structures that can adapt quickly. To me, agentic teams are the practical embodiment of this future: neither fully autonomous machines nor systems that burden humans with every decision—but human–AI teams that think together, learn together, and share responsibility at the same table.
Comment(s)