There is growing evidence that, in the future, many types of work will be done by teams comprising both humans and AI agents. An AI agent is an autonomous software entity that can use tools, access systems, remember context, and take actions with minimal human oversight. Agents are already working alongside humans as dynamic team members who learn, adapt, and continuously optimize.
As these are uncharted waters, organizations and leaders naturally want to know: How will the dynamics of interaction play out? What will work, and what potential weaknesses should be anticipated? How can operations be designed to leverage the strengths of both human and AI workers?
A large-scale experiment conducted on the MindMeld platform provides one of the most detailed studies so far of human-AI collaboration in action. Involving 2,310 participants working on marketing campaign tasks, the study created both human-human and human-AI teams, with randomized assignment. The AI agents used were designed with varying personality traits, and the platform captured detailed logs of communication, task execution, and outputs.
Participants were assigned to teams that either comprised only humans or paired humans with AI agents and were asked to develop ad campaigns using both text and image inputs. This study captured granular logs of team communication, collaboration behavior, and output quality across both lab and field settings. The result: a precise and nuanced view of how AI reshapes teamwork, task allocation, and performance outcomes.
Human-AI Teams Showed 60% Higher Productivity Per Worker
Teams that included AI agents delivered 60% more output per worker than human-only teams. This performance boost was accompanied by a distinct shift in how work was done. Collaborating with AI agents led to a 137% increase in communication and enabled humans to spend 23% more time on generating text and image content through messaging, while spending 20% less time on direct text editing.
Human-AI teams communicated more frequently – sending 45% more messages overall – and focused much more on task-related communication.
The content of these messages was notably different. Human-AI conversations included more planning, prioritization, instructions, and evaluative comments – suggesting a more structured, execution-focused style of working. In contrast, human-only teams exchanged more social and emotional messages, such as rapport-building, self-assessments, and expressions of concern.
This change in communication patterns is not a cosmetic detail and reflects a rebalancing of effort. With AI agents taking on parts of the workload, human participants spent less time managing team dynamics and more time on high-value tasks.
Humans Move Up the Value Chain
In human-AI teams, humans spent 23% more time contributing content, both text and image inputs, and 20% less time editing the AI’s text output. Copy produced by these teams required 84% fewer edits compared to that from human-only teams. This shift allowed human contributors to focus on creative input and overall campaign strategy, rather than line-by-line editing.
These findings suggest that AI agents are not simply assistants and are active contributors that shift how human effort is allocated. When AI agents are capable of producing high-quality outputs autonomously, human collaborators can move up the value chain, taking on roles that involve judgment, ideation, and oversight.
AI’s Strengths Are Not Universal
The study also made it clear that AI’s capabilities are uneven across task types. While human-AI teams consistently outperformed on text-based outputs such as ad copy, they produced lower-rated images than human-only teams. In field tests involving approximately 5 million impressions, the ads that performed best were those combining high-quality human-generated images with AI-assisted text.
This discrepancy is rooted in AI’s current limitations in predicting and optimizing for visual quality. Image generation often relies on aesthetic judgment, contextual relevance, and brand sensitivity – areas where human perception still outperforms current-generation AI models.
For leaders, the takeaway is practical: AI agents should be deployed where they are proven to enhance quality and speed, but human oversight remains essential for outputs involving subjective or brand-sensitive decisions.
Pairing Human and AI Traits Can Influence Output Quality
A particularly innovative aspect of the study involved varying the AI agents’ personalities, such as openness, conscientiousness, and assertiveness, and observing how these traits interacted with those of human collaborators. The findings show that specific combinations led to better outcomes. For example, conscientious humans working with “open” AI agents – those designed to be exploratory, idea-generating, and flexible – produced better image outputs. Conversely, mismatched pairings, such as dominant personalities clashing or both parties being overly passive, resulted in reduced quality.
The ability to design AI behavior (via prompt engineering and agent configuration) opens up a new avenue in organizational design. Just as companies consider skills and working styles when assembling human teams, similar considerations will become relevant in shaping hybrid teams that include AI.
Designing for Hybrid Teams
This research shows that when AI agents are thoughtfully integrated into team structures, they do more than speed up execution. They transform workflows, communication patterns, and the roles that humans play. Designing for this hybrid environment requires deliberate choices in three areas:
- Workflow Design: Anticipate reduced need for social coordination and increased task-focused communication. Workflows should support faster decision-making and seamless handoffs between AI and human agents.
- Role Reconfiguration: Shift human roles toward ideation, oversight, and strategic judgment, especially as AI takes over execution-heavy tasks.
- Agent Configuration: Treat AI personalities as a variable that can enhance or hinder collaboration. Agent design should be matched to the behavioral profiles of human teammates.
AI agents are already becoming embedded in everyday work environments, from co-writing documents to generating reports, proposals, and creative assets. As their autonomy grows, so does the importance of designing systems that bring out the best in both humans and machines.
The time is right to consider your strategic approach to leveraging the power of AI for Sales and Marketing. Let us facilitate a brainstorming workshop for your key stakeholders, and set the ball rolling to design this strategy. Drop us a line at sales@marketaxisconsulting.com to get started.
