COLLABORATING AS A QUINTET: AI AS A TEAMMATE
J.J. (Hans) Schönfeld MCM - Go Flow Network · Go Flow Network: Articles · 2026
Everything is connected. An organization is increasingly unable to solve problems on its own and must learn to collaborate in ecosystems. Today, it is both interesting and important to consider AI as a collaboration partner. This article focuses on three questions: What can AI contribute as a digital collaboration partner? What changes when AI is considered a teammate? What conditions are needed to collaborate responsibly with digital partners such as AI agents?
COLLABORATING AS A QUINTET: AI AS A TEAMMATE
Hans Schönfeld September 2026
AI: from tool to partner
The book Go Flow: fit, snel en dynamisch organiseren in een wereld vol onzekerheid en onkenbaarheid, to be published by Van Haren Publishers on November 1, 2026, describes how boundaries between organizations, roles, and disciplines are becoming fluid. Since ChatGPT's rapid growth following its introduction on November 30, 2022, the boundary between human and artificial intelligence has also quickly become less self-evident. This fluidity affects every aspect of our lives and organizing. Now that AI is available as a digital collaboration partner, the composition of teams and the ways businesses, governments, knowledge institutions, and citizens collaborate in ecosystems are changing (Schönfeld, 2026, in press, expected November 2026).
The effectiveness of teams and ecosystems (collaboration beyond one's own organization) is connected to work design, organizational context, and processes within and outside the team or ecosystem (Cohen & Bailey, 1997). Collaboration requires alignment on the shared challenge, the environment, and the partnership's legal form (Sanders & Heldeweg, 2012). Partners need to understand, use, and trust one another's contributions. Successful team performance and ecosystems require the core ability to know where expertise is available, where it is needed, and how to put it to use. Having knowledge and enabling knowledgeable people to collaborate are therefore two different things (Faraj & Sproull, 2000). The book calls these core capabilities connecting fitness: the ability to organize, maintain, and use institutional integration, or coordinated activities (Schönfeld, 2026).
In the 1990s, Henk Hofstra (an artist from Drachten), Tom van der Goot (then director of the De Lawei theater in Drachten), and I explored alternatives to existing, mostly wet, shaving solutions for Philips. This led to the idea of a waterproof electric shaver that applies a conditioning lotion while shaving in the shower. Philips developed the technology, working with Nivea on the lotion. It became a great success. Philips later did something similar with the Senseo coffee machine, with Douwe Egberts supplying the coffee pods as its collaboration partner. To me, these examples show how value can emerge when technical knowledge, user experience, and creative new perspectives meet. Artificial intelligence played no role in either ecosystem. Yet the example highlights an important question about AI's contribution to collaboration in the AI era.
What additional contributions can AI programs and AI agents make as new participants in partnerships?
The rise of AI requires explicit attention to the effects of human-AI integration in teams and partnerships. AI language models that answer human questions (generative AI programs) have become available at extraordinary speed and are developing into AI agents that perform steps independently. AI can therefore operate as a digital partner, both to an individual and as a teammate. This will undoubtedly influence performance, the speed and quality of production, and how knowledge is created.
Until recently, scientific research mainly viewed AI as a tool for improving individual performance. In my view, that focus created an important blind spot. Large language models are trained on human language and respond through dialogue, just as people do. I therefore expect users to experience them more as social collaboration partners than as mechanical tools (Mollick, 2024). It is both interesting and important to consider AI as a collaboration partner. In this article, I focus on three questions:
- What can AI contribute as a digital collaboration partner?
- What changes when AI is considered a teammate?
- What conditions are needed to collaborate responsibly with digital partners such as AI agents?
What research shows about collaboration
Problems and production processes are becoming more complex. For public challenges, collaboration between government and private parties can help when markets, network governance, or government action alone fall short in serving public interests (Sanders, 2014). Individuals and organizations therefore increasingly work together on products, improvements, and outcomes, sometimes out of necessity and sometimes out of conviction (Deming, 2017; Katz & Lazer, 2003; Puranam, 2018).
Collaboration in teams and ecosystems increasingly forms the backbone of modern organizations.
The Philips example shows how collaboration can integrate different perspectives. Problems can then be approached or solved differently, innovatively, and sometimes more effectively (Ancona & Caldwell, 1992; Cohen & Bailey, 1997; Csaszar, 2012). Joint production also creates organizational challenges: consultation and coordination, for example, inevitably cost time and resources. However, well-structured teamwork can mobilize a broader knowledge base for complex tasks and help achieve goals that individual partners could not reach alone (Alchian & Demsetz, 1972; Cohen & Bailey, 1997).
By sharing the workload and using complementary skills, teams can adapt quickly and dynamically (Go Flow) to changing demands and produce more appropriate outcomes than individuals working alone (Schönfeld, 2026). Delivering a product or addressing a problem generally requires several areas of expertise to work together. Performance depends not only on the expertise available but also on its coordination. Partners must identify where knowledge exists, recognize where it is needed, and coordinate its use (Beane & Anthony, 2024; Faraj & Sproull, 2000; Orlikowski, 2002). Working with others also provides social and motivational benefits that increase job satisfaction compared with working alone (Deutsch, 1949; Johnson & Johnson, 2005; Kozlowski & Bell, 2013).
The effects of AI as a digital collaboration partner
Let us now turn to AI programs and AI agents as collaboration partners. Because ChatGPT only introduced AI to the wider public in late 2022, the available studies are recent. They demonstrate better individual creative performance with AI in controlled settings (Dell’Acqua et al., 2023; Noy & Zhang, 2023). Until recently, however, no reliable study had examined AI's effects in collaborating teams. Important knowledge was added in 2026 (Dell’Acqua et al., 2026). Amid the continuing stream of negative reports about unspecified disruptive forces attributed to AI (we shall see), this research provides a positive counterweight.
AI's effects on teamwork were examined in a large experiment involving 791 Procter & Gamble professionals. Participants were randomly assigned to work on challenges with or without AI: an individual alone and an individual supported by an AI program (groups A and B), and two people in a team compared with two people supported by an AI program (groups C and D). The results offer interesting insights into generative AI's influence within organizations and teams (Dell’Acqua et al., 2026).
AI improves solution quality
The effect is substantial. Individuals with AI (group B), for example, achieved average solution quality comparable to that of a two-person team without AI (group C). This means collaboration with AI makes a major contribution to quality. An individual working with AI gains a significant portion of the benefit that would otherwise come from working with a second person (Dell’Acqua et al., 2026).
AI makes proposals less one-sided
Without AI (groups A and C), proposed solutions often reflect employees' backgrounds. Research and development employees, for example, emphasize technical aspects, while proposals from employees with commercial backgrounds focus more on the market. The study found that individuals and teams working with AI (groups B and D) produced substantially more balanced proposals: broader and less dependent on personal background.
This suggests that AI helps individuals and teams look beyond their own field and incorporate other aspects into a product (Dell’Acqua et al., 2026). AI appears able to take on certain collaborative functions by extending users' reach beyond their core expertise. Employees use AI suggestions in practice to bridge knowledge gaps. This reproduces part of the cross-functional knowledge integration normally achieved through human collaboration.
Functional boundaries will gradually blur as expertise becomes more accessible.
AI makes work more enjoyable
Employees working with AI (groups B and D) reported more positive and fewer negative emotions at work than participants working alone. This supports my view that AI will be experienced less as an instrument and more as a person, potentially fulfilling some of the emotional and motivational functions of collaboration (Dell’Acqua et al., 2026).
Human judgment remains necessary
AI's greatest contribution was to the quality of generated ideas. Quality in groups B and D was substantially better than in A and C. That contribution did not, however, improve the ability to select the best idea from a set of good ones. Human judgment proved most valuable there. In this study, AI emerges as a digital partner that strengthens quality and improves the raw material for innovation, while human judgment is better at recognizing the promising concepts within it (Dell’Acqua et al., 2026).
From quadruple helix to quintet
The quadruple helix helps us consider collaboration beyond a single team. In brief, this model connects knowledge institutions, businesses, and government with the public and society (Carayannis & Campbell, 2009). In practice, we often speak of collaboration among businesses, governments, knowledge institutions, and citizens.
The book Go Flow, fit, snel en dynamisch organiseren describes the fluid transition from working for people, such as citizens, to working with them. A striking example is the international Joint Investigation Team (JIT) working with the Bellingcat collective of citizens and journalists in the investigation of the MH17 disaster. The government team collaborated with businesses, scientists, and citizens. Bellingcat's findings contributed substantially to the evidence and helped make convictions possible in the eventual trial.
Building on quadruple helix collaboration, the new book introduces collaboration as a quintet by adding a fifth partner: intelligent systems that independently contribute substantively to analysis, idea generation, and execution (Schönfeld, 2026). By quintet, I mean a form of collaboration in which AI has a recognizable substantive and operational role. The human and institutional participants, who may themselves also use AI, define the assignment, limit the scope for action, and remain responsible.
Figure 1: Collaboration as a quintet (Schönfeld)
AI becomes a partner that can independently make proposals that influence choices. The question is how we can and will organize this collaboration in practice. The P&G research offers empirical starting points for organizing with digital partners at the team level. Further research should distinguish between an AI language model that responds to questions and a digital AI agent that is authorized and able to carry out several steps within an assignment. I am particularly referring to AI agents: systems given a defined goal, tools, and a certain scope for action.
How independently an AI agent can and may work depends on its design, its task, and the access it is given or permitted to use.
The P&G study was limited to collaboration with generative AI. In my view, further research should examine situations in which AI agents, as digital partners, can monitor progress, gather information, or prepare or perform follow-up actions. In Go Flow, I explore several roles in organizing with AI agents and cobots (Schönfeld, 2026), considering possible applications that will need to be tested in each situation.
Conditions for working as a quintet
As we allow digital partners to perform more work in partnerships, the agreements governing that work become more important. What information may a digital partner use? What actions may it take independently? Who checks quality, and when must the digital partner stop or return with a question?
Research on collaboration without digital partners shows that effectiveness and efficiency must be balanced with legitimacy and legal validity (Sanders & Heldeweg, 2012). This should also, perhaps especially, apply to teams involving digital partners. It raises a question for further research into the collaboration quintet: who may decide, within what limits, and under whose responsibility?
Recent incidents reported in the news show that major American technology companies currently allow their systems too much freedom, perhaps influenced by geopolitical considerations. Findings from investigations into these incidents are unfortunately not always public, and other incidents may remain outside public view. Sometimes only anonymized summaries of incidents and causes are available, and sometimes nothing is available at all. That is regrettable. Research into such incidents should clarify the importance of clear boundaries and oversight. Openness about causes and measures is needed to learn from these experiences. For me, digital fitness for people, teams, and ecosystems has a very practical meaning. As I write in Go Flow:
“Digital fitness here means being able to understand what you hand over and what you remain responsible for” (Schönfeld, 2026, p. 110).
Give AI a recognizable role. For one challenge, that may mean suggesting alternatives; for another, comparing arguments or preparing a shared overview. An AI agent can act as a partner to carry out follow-up steps, within what I believe should be clearly defined, built-in limits established in advance. The finding that AI primarily improved idea generation makes deliberate attention to selection important. For now, leave that selection to a human (Dell’Acqua et al., 2026).
Leaders who organize room to move
In Go Flow, leaders are described as choreographers who create conditions in which people and systems can move together, improvise, and adjust their actions in small steps (Schönfeld, 2026). In this way, leaders organize the space in which partners, including digital ones, may contribute and the moments when those contributions must be reassessed.
The collaboration quintet is promising in terms of potential outcomes and invites practical research into digital partners' effects in ecosystems. What do they add, what connections emerge, and what does this require of the five types of partners? The Go Flow approach focuses on the ability to keep acting appropriately while reality continuously shifts. This includes the movement toward using, questioning, constraining, and, when necessary, reorganizing AI as digital partners.
Fit collaboration means not fixing our division of roles around what is normal today, but also looking at what is possible.
References
Alchian, A. A., & Demsetz, H. (1972). Production, information costs, and economic organization. The American Economic Review, 62(5), 777–795. Ancona, D. G., & Caldwell, D. F. (1992). Bridging the boundary: External activity and performance in organizational teams. Administrative Science Quarterly, 37(4), 634–665. https://doi.org/10.2307/2393475 Beane, M., & Anthony, C. (2024). Inverted apprenticeship: How senior occupational members develop practical expertise and preserve their position when new technologies arrive. Organization Science, 35(2), 405–431. https://doi.org/10.1287/orsc.2023.1688 Carayannis, E. G., & Campbell, D. F. J. (2009). “Mode 3” and “Quadruple Helix”: Toward a 21st century fractal innovation ecosystem. International Journal of Technology Management, 46(3/4), 201–234. https://doi.org/10.1504/IJTM.2009.023374 Cohen, S. G., & Bailey, D. E. (1997). What makes teams work: Group effectiveness research from the shop floor to the executive suite. Journal of Management, 23(3), 239–290. https://doi.org/10.1016/S0149-2063(97)90034-9 Csaszar, F. A. (2012). Organizational structure as a determinant of performance: Evidence from mutual funds. Strategic Management Journal, 33(6), 611–632. https://doi.org/10.1002/smj.1969 Dell’Acqua, F., Ayoubi, C., Lifshitz, H., Sadun, R., Mollick, E., Mollick, L., Han, Y., Goldman, J., Nair, H., Taub, S., & Lakhani, K. R. (2026). The cybernetic teammate: A field experiment on generative AI and teamwork. Organization Science, 37(4), 1217–1242. https://doi.org/10.1287/orsc.2025.20702 Dell’Acqua, F., McFowland, E., III, Mollick, E. R., Lifshitz-Assaf, H., Kellogg, K. C., Rajendran, S., Krayer, L., Candelon, F., & Lakhani, K. R. (2023). Navigating the jagged technological frontier: Field experimental evidence of the effects of AI on knowledge worker productivity and quality (Working Paper 24-013). Harvard Business School. https://www.hbs.edu/ris/download.aspx?name=24-013.pdf Deming, D. J. (2017). The growing importance of social skills in the labor market. The Quarterly Journal of Economics, 132(4), 1593–1640. https://doi.org/10.1093/qje/qjx022 Deutsch, M. (1949). A theory of co-operation and competition. Human Relations, 2(2), 129–152. https://doi.org/10.1177/001872674900200204 Faraj, S., & Sproull, L. (2000). Coordinating expertise in software development teams. Management Science, 46(12), 1554–1568. https://doi.org/10.1287/mnsc.46.12.1554.12072 Johnson, D. W., & Johnson, R. T. (2005). New developments in social interdependence theory. Genetic, Social, and General Psychology Monographs, 131(4), 285–358. https://doi.org/10.3200/MONO.131.4.285-358 Katz, N., & Lazer, D. (2003). Building effective intra-organizational networks: The role of teams [Working paper]. Center for Public Leadership, Harvard University. https://hdl.handle.net/1721.1/55801 Kozlowski, S. W. J., & Bell, B. S. (2013). Work groups and teams in organizations: Review update. In N. Schmitt & S. Highhouse (Eds.), Handbook of psychology: Vol. 12. Industrial and organizational psychology (2nd edition, pp. 412–469). Wiley. https://doi.org/10.1002/9781118133880.hop212017 Mollick, E. (2024). Co-intelligence: Living and working with AI. WH Allen. Noy, S., & Zhang, W. (2023). Experimental evidence on the productivity effects of generative artificial intelligence. Science, 381(6654), 187–192. https://doi.org/10.1126/science.adh2586 Orlikowski, W. J. (2002). Knowing in practice: Enacting a collective capability in distributed organizing. Organization Science, 13(3), 249–273. https://doi.org/10.1287/orsc.13.3.249.2776 Puranam, P. (2018). The microstructure of organizations. Oxford University Press. https://doi.org/10.1093/oso/9780199672363.001.0001 Sanders, M. (2014). Publiek-private samenwerking: Een reparatiestrategie voor falende ordeningsvormen. Bestuurskunde, 23(3), 69–77. Sanders, M., & Heldeweg, M. A. (2012). Publiek-private samenwerking: Evenwichtskunst tussen juridisch-bestuurskundige waarden. Bestuurswetenschappen, 66(1), 39–56. Schönfeld, J. (in press, expected November 2026). Go Flow, fit, snel en dynamisch organiseren: In een wereld vol onzekerheid en onkenbaarheid. Van Haren Publishing.
Related publications
Univé Rechtshulp rolls out legal AI support with LegalMike
LegalMike · LegalMike · May 19, 2026
Univé Rechtshulp is connecting its internal AI assistant to LegalMike’s legal research platform so its lawyers can spend more time with clients.
LegalMike aims to be a legal research partner
Kees Pijnappels · Advocatenblad · October 8, 2025
LegalMike brings Dutch and European legal sources together so lawyers can check the authority behind a proposed answer.
Dutch legal aid clinics gain free access to a legal AI assistant
LegalMike · LegalMike · September 11, 2025
LegalMike is making its research assistant available free of charge to Dutch legal aid clinics as access to legal help comes under pressure.
Discussions about this
Have a question about this topic, or an experience to share? Start a discussion; it stays linked to this page.
Comments
No comments yet
Be the first to comment.
Want to comment? Log in or join for free.