While artificial intelligence is now omnipresent in technological debates, few organizations still manage to fully integrate it into their business processes. The majority of companies, despite rapidly increasing investments, remain at the stage of experiments or pilot projects. It is in this context that Kyndryl, a major player in technology services, reveals a fundamental breakthrough with agentic AI, a significant evolution towards autonomous intelligence capable of acting and no longer just responding. This innovation promises to revolutionize working methods, productivity, and engagement by introducing agents capable of carrying out complete chains of actions with unprecedented autonomy. Habib Messaoudi, Vice President Cloud, Applications, Data & AI at Kyndryl France, shares in this exclusive interview his analysis of the current barriers to large-scale AI deployment, the transformations of professions to come, and how agentic AI opens a new era for enterprises.
- 1 Persistent obstacles to scaling artificial intelligence projects
- 2 Clearly defining agentic AI: an active and autonomous intelligence
- 3 Profound transformations of professions and skills related to the rise of agentic AI
- 4 Revolution of workflows and IT architectures induced by the integration of AI agents
- 5 Concrete measures of productivity gains linked to agentic AI
- 6 Human and organizational challenges to overcome for successful adoption of agentic AI
- 7 Future perspectives and evolutions of agentic artificial intelligence
- 8 FAQ on agentic AI and its impacts in organizations
Persistent obstacles to scaling artificial intelligence projects
For several years, the technological landscape has been marked by a multiplication of initiatives around artificial intelligence. Yet, despite growing interest and investments rising by about 33% year-on-year according to the Kyndryl Readiness Report 2025, the transition from pilot stage to operational integration remains a major challenge. Nearly 62% of organizations continue to explore this technology without managing to sustainably deploy applications at large scale.
This dichotomy highlights deep structural mismatches. For example, although 90% of leaders acknowledge that their innovative tools foster creativity, a majority point out that their technological infrastructure struggles to keep pace. In France, 57% of decision-makers consider their legacy systems more a hindrance than a support to innovation.
The primary obstacle often lies in the complexity of infrastructures: old systems, poorly designed or siloed architectures, and above all fragmented governance with poorly integrated security measures. This observation is not just a technical problem but also a human one. The required skills remain insufficient and poorly aligned with technological evolutions.
Thus, modernizing tools alone without simultaneously training and engaging talents creates an illusion of progress, while human and professional capacities to fully exploit artificial intelligence are still seriously lacking. AI remains confined to experimental demonstrations, unable to generate a real impact at the heart of business processes.
This situation is illustrated by many cases where an ambitious project ends as a simple proof of concept due to lack of cross-functional team support, while the technology would have the potential to sustainably transform the company. It is precisely in this context that agentic AI comes into play, presenting an innovative solution to overcome this hurdle and reconcile technical capabilities with organizational adoption.
Clearly defining agentic AI: an active and autonomous intelligence
At the heart of this technological revolution lies agentic AI, a concept that goes beyond the traditional capabilities of artificial intelligence. Unlike generative AIs, which are limited to producing content or punctual responses based on human queries, agentic AI plays a proactive role, capable of planning, deciding, and performing series of actions without continuous intervention.
To better grasp this advancement, it is useful to distinguish three major categories of artificial intelligence:
- Traditional AI: analyzes and predicts from data, facilitating decision-making by humans.
- Generative AI: creates content in response to a specific request, such as texts, images, or sounds.
- Agentic AI: acts autonomously by interacting directly with systems, orchestrating complex end-to-end tasks.
Agentic AI thus embodies active intelligence, capable of contextual adaptation within a framework of rigorous governance and security. Kyndryl, a pioneer in the race for this technology, has set up the Kyndryl Agentic AI Framework, a system designed to efficiently deploy autonomous agents in existing infrastructures, ensuring integration, security, and scalability.
This advancement marks a decisive step in the maturation of AI towards concrete professional uses, capable of automating entire processes and significantly boosting productivity.
Operation and benefits of agentic AI in daily work
An intelligent agent, endowed with learning and adaptation capabilities, can for instance automatically manage technical incidents, make corrective decisions after analysis, or manage complex workflows without ever losing track of an evolving situation.
By combining automation and intelligence, this type of AI reduces human errors, accelerates response times, and frees employees from repetitive tasks. For example, in the financial sector, autonomous AI agents can handle credit requests with unparalleled efficiency, ensuring both compliance and responsiveness.
Another major advantage is the possibility to query these agents in natural language while benefiting from clear explanations of their decisions. This enhanced transparency helps establish a climate of trust that facilitates faster and broader adoption.
The arrival of agentic AI promises to durably disrupt roles and responsibilities within organizations. According to indicators highlighted by the Kyndryl Readiness Report 2025, nearly 90% of decision-makers anticipate a major reconfiguration of their teams.
However, this is not a substitution but rather mutual enrichment. AI agents take over low added-value operations, allowing employees to focus on strategic tasks such as management, creativity, or handling complex interactions. This coexistence marks the advent of a hybrid workforce where humans and autonomous agents collaborate effectively.
To succeed in this transition, training and supporting teams are indispensable to develop skills adapted to these new environments. Continuous learning, mastering AI tools, and understanding governance mechanisms become pillars to maximize the generated value.
A concrete example concerns IT support professions. Autonomous AI agents can automatically handle incidents following protocols, while technicians can deepen complex diagnostics or process improvements. This paradigm shift leads to qualitative skill upgrading.
The human challenge at the heart of the agentic AI revolution
The main challenge does not rest solely on technology but on the willingness to adopt new modes of collaboration and decision-making. A cultural change is often necessary, as nearly half of leaders still lament a lack of agility in their organization, hindering innovation.
Transparency and trust play a fundamental role in the appropriation of these autonomous agents. By placing humans at the center of the decision-making loop, while harnessing the strengths of automation, companies can co-construct a future where agentic artificial intelligence is a lever for sustainable innovation.
Revolution of workflows and IT architectures induced by the integration of AI agents
The effective integration of intelligent agents into professions requires a redesign of processes and system architectures. Workflows must become seamless, thus avoiding the multiplication of silos that would slow synergy between different services or technologies.
Moreover, the robustness of architectures is paramount. They must be able to absorb the load increase inherent in the massive deployment of AI agents. Kyndryl notably relies on a resilient infrastructure and a “zero trust” governance model, guaranteeing that every interaction is secure and traceable.
Systems must also promote interoperability with existing business applications to ensure native integration. The Kyndryl Agentic AI Framework is thus based on advanced design principles to secure, orchestrate, and manage agents in critical environments, such as mainframes, where business continuity is crucial.
This framework also avoids tedious duplication of developments and facilitates the rapid scaling up of multiple autonomous agents, generating significant operational benefits.
Key elements for successful deployment of AI agents in the company
- Smooth integration to avoid silos and ensure data continuity.
- Resilient architecture capable of handling heavy loads and potential interruptions.
- Security built-in from design with zero trust strategy and complete traceability.
- Interoperability with existing systems and capacity for future evolutions.
- Transparent governance and human controls to maintain trust and compliance.
Concrete measures of productivity gains linked to agentic AI
Early deployments of AI agents show tangible results, going beyond the experimental framework. The Kyndryl Readiness Report 2025 reveals that 54% of companies observe a positive return on investment, often on targeted use cases combining efficiency and cost reduction.
In the insurance field, the adoption of autonomous agents reduces by 40% the time required for claim processing, while losses related to fraud decrease between 25 and 30%. These figures illustrate a very concrete impact, allowing companies to improve their competitiveness.
In finance, gains are just as remarkable, with decreases up to 60% in case processing durations, increased visibility by 90% on risks, and a 60% reduction in manual work. These results are not isolated but reflect a strong trend towards accelerated intelligent automation at large scale.
| Sector | Productivity Gain | Specific Improvement |
|---|---|---|
| Insurance | -40% claim processing time | -25 to -30% losses related to fraud |
| Finance | -60% processing time | +90% risk visibility |
| IT / Support | -50% human intervention | Automation of recurring incidents |
These successes demonstrate that when infrastructures, security, and governance are properly put in place, agentic AI can surpass the experimental stage to produce measurable and lasting effects.
Human and organizational challenges to overcome for successful adoption of agentic AI
Despite these technological advances, France shows some delay in preparing teams to integrate this new form of intelligence. Only 29% of decision-makers believe their employees master the necessary skills to work with autonomous agents.
This challenge therefore goes beyond the simple technical question. It relies on the ability of organizations to evolve culturally and managerially.
Kyndryl identifies three major levers to succeed in this transformation:
- Hybrid approach: placing humans at the center of the decision loop to guarantee transparency, control, and trust in AI agents.
- Cultural change: improving organizational agility to facilitate adoption of new working methods and accelerate decision-making processes.
- Explainable and responsible systems: developing AI that is understandable, traceable, and compliant with ethical requirements to establish lasting trust.
It is only under these conditions that the revolution propelled by agentic AI will be able to unleash its full industrial-scale potential, transforming companies into agile, innovative, and resilient actors.
Future perspectives and evolutions of agentic artificial intelligence
While some analysts like Gartner have expressed reservations by labeling agentic AI as a fad likely to fade after 2027, early feedback from the field suggests instead rapid maturity and growing adoption.
The approach proposed by Kyndryl through its Agentic AI Framework paves the way for hybrid environments where humans and machines collaborate harmoniously. This symbiosis should improve not only productivity but also the innovation capacity of companies.
The immediate future of agentic AI also points toward better integration in critical sectors such as healthcare, industry, or defense, where the need for autonomy, precision, and security is paramount.
One can anticipate the emergence of new economic models and deep organizational changes that will rely on this symbiosis between intelligent agents automating heavy tasks and humans bringing meaning, creativity, and oversight.
This advancement opens unprecedented prospects where AI ceases to be a mere tool to become truly a partner in the success of organizations.
FAQ on agentic AI and its impacts in organizations
What is agentic AI?
Agentic AI is a form of artificial intelligence capable of making autonomous decisions and executing complex tasks without constant human intervention, going beyond simple responses or content creation.
How does agentic AI transform professions?
It automates repetitive and operational actions, allowing employees to focus on higher value-added tasks such as innovation, supervision, or decision-making.
What are the main barriers to the adoption of agentic AI?
Major obstacles are related to aging infrastructures, insufficient human skills, and a lack of organizational agility in decision-making processes.
What concrete benefits can companies expect?
According to early feedback, gains include significant reductions in processing times, better risk management, and decreases in errors and fraud, leading to a notable improvement in productivity.
How to prepare teams to work with intelligent agents?
It is essential to invest in continuous training and to promote a cultural change that places humans in the decision-making loop, while encouraging trust in the technologies used.
