EN
EN DE

Tag: AI

  • KI ist kein Teamkollege. Aber das beste Tool für dein Team.

    AI is not your teammate. But the best tool for your team.

    Everybody is building AI agents nowadays. The future of work seems to be Humans und AI working together in teams. Probably it’s a natural human tendency - we create in our own image. But while we are trying to understand the workforce of tomorrow, I'm increasingly convinced that framing artificial intelligence as "teammates" misses an important point.

    My leadership mission has always been straightforward: help people build better products while making work less miserable. Throughout my career, I've observed how easily our work environments can become disconnected from what makes us fundamentally human. Companies need innovation, people crave meaningful contribution yet work often feels hollow. Addressing this problem was a fundamental concern of the agile movement.

    Today AI transforms our professional landscape. Undoubtedly it boosts agility through faster feedback loops and smaller teams of AI-augmented professionals. Yet it might put the human-centric view on work that is at the core of the agile mindset at risk. So should we strive for human-like coworkers, or should AI be envisioned as something uniquely complementary to human nature? Do we really want to go for lunch with AI teammates?

    Replace or Enhance

    In a recent article, Jurgen Appelo outlined his vision for "After Agile"—a world where AI assistants become interoperable tools and teammates, speaking a new "language" of protocols, semantics, negotiation, and auditability. He envisions systems to manage relationships, operations, and negotiations between people and autonomous agents.

    Every power user of AI-tools will agree that Jurgen's call to invent new protocols for human-agent collaboration is compelling. Today, we are massively limited by our ability to work with AI, and the answer to this is certainly not Prompt Engineering. But I wonder if we're asking the right question here. Instead of "how do we make AI more like human colleagues?" perhaps we should ask: "how do we leverage AI to enhance what makes us uniquely human?"

    The Trap of Human-like AI

    Many companies are racing to build AI personal assistants—digital entities designed to simulate human collaboration. The underlying assumption is that these agents should become increasingly human-like, eventually functioning as "real" coworkers.

    This approach seems futuristic yet intuitive, but I believe it omits fundamental aspects of both human psychology and our relationship with technology. To chart an alternative course, we need to recognize and harness three uniquely human superpowers: tool-use, collaboration and adaptation.

    Human Superpower #1: Tools

    An evolutionary advantage as human beings is our unparalleled capacity to incorporate tools as extensions of ourselves. From hand-axes to smartphones, human brains literally rewire themselves to treat tools as body extensions.

    Think of the bicycle as metaphor. For a professional cyclist, bike and body exist in perfect symbiosis - every muscle firing, every gear shift intuitive. There's no conscious translation between intent and execution. More importantly, this mastery brings profound joy. The harder the climb, the greater the thrill at reaching the summit.

    AI agents can serve as next-generation tools. They can eliminate repetitive tasks, but their true potential lies in becoming "cognitive prosthetics" that amplify our curiosity, creativity, and problem-solving stamina. Try using deep research to fact-check an idea, without days of market research. Our own ability to think critically is a scarce resource that now can be extended by AI.

    Human Superpower #2: Collaboration

    Our second superpower is genuine human connection. When we collaborate effectively, our brains release oxytocin and serotonin—creating that unmistakable "warm glow" of working together. Shared purpose and trust spark creativity and well-being in ways no algorithm can replicate. While AI can automate drudgery, it cannot trigger a genuine spark of trust or spontaneous insight born from human connection.

    Our brains evolved sophisticated capabilities to distinguish friend from foe through micro-expressions, vocal tones, and body language. Our ability in detecting subtle human cues means AI will always fall short of real human interaction. These detection systems operate below conscious awareness, making it impossible for AI to fully satisfy our need for genuine human connection.

    It’s not even their fault - even if the technology will become almost indistinguishable from Humans, we will still know that it’s not like us. Thus, treating agents as full teammates can hollow out genuine community and connectedness, potentially increasing workplace loneliness. We might achieve reach higher levels of productivity, but at the price of misery.

    Human Superpower #3: Adaptation

    Rather than forcing AI into human-like roles, I propose a different approach that builds on a third superpower: our ability to adapt. Across history, humans have continuously reshaped how we live and work in response to new technologies - whether inventing agriculture, building cities, or navigating the digital age. Each shift brought up new tools and new ways of collaborating, organizing, and thinking.

    In this vision, agents serve as augmenters, not replacements. They accelerate outcomes and expand our cognitive and creative bandwidth. But human teams remain central. We design workflows where AI boosts our performance while we retain both the joy of mastering tools and the warmth of collaboration.

    Of course, we still need more effective ways to work with AI, and I fully agree with Jurgen’s call for a language that goes beyond prompting. AI augmentation will have tremendous impact on human ability to perform and succeed. This can transform work from tedious to thrilling and bring joy to teams. But the emotional aspects of work, the setbacks, the breakthroughs, the thrill of succeeding together, will always be uniquely human.

    The future of work isn't about building better artificial colleagues. It's about creating systems where technology amplifies what makes us most human: our ability to extend ourselves through tools and connect deeply with one another. As leaders, our challenge isn't teaching machines to be more human. It's designing environments where technology helps humans be more fully themselves: creative, connected, and engaged in meaningful work.

  • Algorithmisches Selbst-Management

    Algorithmic Self-Management

    AI might just solve one of the most challenging human tasks in management

    A good friend recently quit his job—and I've rarely felt so relieved. For months, I'd watched as he struggled under a barrage of feedback from his managers. Each piece of advice was well-intentioned, designed to "help him grow." Instead, it systematically broke down his confidence and undermined his performance. Ironically, the very feedback intended to elevate him nearly drove him into depression.

    This isn't an isolated case. Feedback, as practiced in many workplaces today, is fundamentally broken. Leaders feel obligated to point out flaws, believing it's their duty to guide employees toward improvement. Employees, meanwhile, are expected to accept criticism with humility, or risk being labeled "uncoachable." But what if we've misunderstood the whole concept of feedback?

    Feedback doesn't simply fail because people dislike criticism. It fails because it's inherently subjective and emotionally damaging. We've been conditioned to believe feedback is indispensable. Yet, the evidence—and our lived experience—suggests it doesn't work. It's time we rethink how we support people in their growth. Clearly, feedback isn't the answer.

    The Science Against Feedback

    Marcus Buckingham und Ashley Goodall conclude: Human judgment is deeply flawed. When managers evaluate performance, their assessments are overwhelmingly influenced by their own biases and perceptions—what researchers call the idiosyncratic rater effect. The result? Feedback becomes more reflective of the evaluator than of the person receiving it.

    Goodhart's Law further compounds this issue, reminding us that as soon as a measure becomes a target, it ceases to be a useful measure. When teams sense that metrics are being used against them, they start gaming the system, rendering feedback even less reliable and effective.

    Between the Stopwatch and the Void

    Historically, management approaches swung between extremes. Taylorism famously optimized productivity in industrial factories through rigid, top-down management—but it collapsed when faced with the complexity and creativity required in knowledge work. Agile methods emerged to address this complexity, promising autonomy and self-organization. Yet, too often, Agile leadership dissolved into a laissez-faire approach, creating a leadership void and leaving teams stranded without clear direction or accountability.

    Meanwhile, companies like Uber resurrected Taylorist principles, wielding algorithms to manage low-skilled workers. This "algorithmic management" was efficient but dehumanizing, as workers experienced relentless monitoring without autonomy or trust.

    People don't want to be micromanaged by inflexible algorithms, nor abandoned by hands-off managers. They crave autonomy but still need structured support to succeed. This paradox sets the stage for a new approach—algorithmic self-management.

    The Clock Doesn’t Lie

    Elite athletes measure themselves relentlessly against the clock, not because the clock judges them, but because it provides clear, unbiased, immediate feedback. It doesn’t criticize, nor does it flatter—it simply tells the truth. Athletes don’t blame the clock when they underperform; they internalize responsibility, adjust their approach, and strive to improve. In complex, collaborative environments like tech teams, an unbiased, data-driven measure can serve a similar purpose. Teams need a "stopwatch" that provides clear, unemotional insights into their performance. But rather than top-down monitoring, this "clock" should empower teams to manage themselves, fostering continuous improvement and mastery.

    Self-Management powered by AI

    Algorithmic self-management powered by empathetic AI offers teams the best of both worlds: autonomy and structure. This method combines the findings of modern labor research and algorithmic management as described by Jurgen Appelo in his book „Human Robot Agent“ where automated systems take over management functions. Imagine a smart assistant integrated seamlessly into the workflow—observing processes, analyzing data from user stories, velocity, code quality, deployments, and bug resolutions. This AI doesn't pass judgment; it offers observations and asks insightful questions designed to spark team-driven improvements.

    Unlike human managers, this AI can process vast amounts of data objectively and consistently, detecting patterns and anomalies in real-time. Its suggestions are neither punitive nor personal. They're designed purely to enhance effectiveness, efficiency, and ultimately, mastery. Engineers receive targeted insights into their performance, empowering them to adjust their practices proactively rather than reactively.

    Radical Candor, Reimagined

    For AI-driven self-management to thrive, organizational culture must embrace Radical Candor, reimagined. Trust and transparency must underpin every interaction. Leadership must commit unwaveringly to the philosophy that AI analytics are a tool for growth, not surveillance. The goal isn't to police teams but to empower them.

    This framework depends entirely on genuine commitment from both leaders and team members. Leaders must protect the integrity and impartiality of the AI, resisting any temptation to misuse or weaponize the data. In return, teams must embrace self-accountability, using AI-generated insights as the trusted, neutral voice that encourages growth without the interpersonal friction of traditional feedback. Algorithmic self-management represents a transformative shift toward true autonomy, mastery, and continuous improvement—one that finally delivers on the promise of agile principles.

    One Step Further: A Glimpse Into the Future

    Imagine a product team struggling to deliver an MVP on time. They're committed, talented, but consistently fall short. Rather than bringing in a manager to deliver tough feedback, the team consults their AI-powered dashboard. The AI identifies that their current velocity isn't sufficient to meet the product milestones. It highlights key bottlenecks, suggests backlog refinements, encourages pair programming, and even recommends specific skill development activities. Building on the recent advances in artificial empathy it might be able to read the room, catch social cues, and deliver insights in most suitable way.

    The team engages openly with these insights, free from the interpersonal tensions typically associated with critical feedback. With clear, actionable advice from an impartial source, they quickly adapt their strategy, regain momentum, and hit their targets without any blame, defensiveness, or damaged relationships.

    Let the Stopwatch Talk

    Traditional feedback has failed us, and outdated management approaches have proven insufficient in today's complex work environments. The solution isn't abandoning structure or retreating into rigid, impersonal algorithms. Instead, we need a smarter, more humane approach—algorithmic self-management powered by empathetic AI.

    Let's give teams the stopwatch they deserve: clear, unbiased, supportive, and always empowering. It's time we let the stopwatch talk and help every team achieve its potential.

    I'd love to hear your perspectives - can AI transform how we manage performance by encouraging autonomy and mastery instead of traditional feedback? Talk to us.