Innovation Management in the AI Future: How humans and machines complement each other

I’m convinced: AI agents will shape the future – across all industries. The technological barriers to automation are rapidly decreasing, opening up entirely new efficiency potentials. That’s why I can’t imagine a better topic to kick off my new series “neosfer Innovation Briefing – with Kai Werner.” In the years to come, corporate innovation units will increasingly focus on experimenting with AI agents and gradually putting them to use – including us.

Just a year ago, AI was still seen by many as a smart assistant: some text generation here, some meeting summaries there – useful, but limited. But what I’ve seen in recent months has fundamentally changed my understanding of AI – and the AI future.

We’re no longer talking about tools – we’re talking about agents. Intelligent software units that take on specific tasks: researching information, making decisions, and carrying out actions. Autonomously, without a human constantly looking over their shoulder.

These agents fall into three categories:

For those who want to dive deeper: my colleague Bala Nagaraj recently summarized the technical details and challenges in a great read.

At Google Next last month in Las Vegas, the tech giant demonstrated not only how far they’ve come in AI – but also the real-world applications already in place.

Some examples:

• Rethinking customer service: A telecom company used AI agents to reduce its average support time by 40%. The agents analyze requests, suggest solutions – and free up employees for more complex tasks.
• Back-office automation: In the insurance industry, AI agents help analyze claims and generate recommendations. The Commerzbank also shared a use case: agents automatically process tasks like address changes – with no manual steps.
• Smarter customer interaction: A global retailer uses AI agents in its customer communication. The result: faster response times and significantly higher conversion rates.

What impressed me most was an AI-supported credit consultation: In a compelling demo, a payslip and personal criteria were submitted – and within seconds, real estate offers and tailored financing options appeared. No handoffs, no waiting, no friction. Behind the scenes: a finely orchestrated interplay of agents demonstrating how far automation has already come.

What’s currently changing in banking operations can also be observed in venture building. Generative AI significantly lowers the threshold for developing new products and business models – and fundamentally reshapes how innovation units work.

Whereas business ideas used to be developed over weeks of workshops and refined with consultants, today AI-powered tools allow for much faster exploration and validation of concepts. Business model drafts, market analyses, or prototypes can now be created in hours. Tools like Bizway, Notion AI, or Figma with AI plugins enable business plans, MVPs, and user testing setups with just a few clicks.

That doesn’t mean launching a successful startup is suddenly easy. But the process is more accessible, faster to test – and more iterative. For innovation units like neosfer, that means: we can pursue more ideas simultaneously, validate hypotheses faster, and allocate resources more strategically. We continue to learn through iteration – just faster and more efficiently. In short: AI becomes a tool that significantly enhances how we approach venture building.

As impressive as these examples are: AI does not take responsibility. It prepares, analyzes, suggests. But the decisions – they must still be made by us humans.

Especially in a highly regulated industry like ours, the principle of human-in-the-loop is essential: only when AI agents operate within clearly defined boundaries and human oversight is guaranteed can trust emerge. It requires transparency, control – and a willingness to collaborate.

A clear division of labor is critical: machines should primarily support repetitive, rule-based, or tedious decisions. This gives us humans the freedom to focus on what requires empathy, context, or ethical judgment.

At Google Next, it became clear where the journey is heading in the AI future – and the role AgentSpace can play. With AgentSpace, Google introduced a platform that allows companies to build, manage, and scale their own agents – some even without coding knowledge. Specialized agents for deep research or idea generation are also available.

What this means: a democratization of AI innovation. Where technical barriers once slowed progress, platforms like AgentSpace could now act as accelerators – even in large enterprises and banks.

At the same time, these developments are fundamentally changing internal workflows in banks. No-code and low-code platforms increasingly enable employees in business units to automate their own processes. They need little or no support from developer teams. As a result, business units are no longer just executing processes – they are actively shaping them.

This shift is also changing the role of IT departments. Thanks to more efficient tools, smaller processes that were previously not worth automating are now being tackled. Developers are no longer working on a single project with one product owner – but simultaneously with multiple business experts, each acting as the owner of their own process. This requires new collaboration models, more communication – and a deep understanding of the business domain.

The question is no longer whether we will adopt AI agents – but how quickly. The AI future is already here. And one thing is clear to me: those who ignore the potential of this new technology risk falling behind. Those who embrace it strategically will gain a real competitive edge.