70 of% initiatives Data Mesh fail in the first two years. The conclusion is clear: this approach, meant to bring agility and autonomy, often clashes with on-the-ground realities.
But a new generation of tools, the Artificial intelligence (AI) agents, could well be a game-changer.
Today we are witnessing the meeting of two major transformations:
Together, these two approaches pave the way for an enterprise where data is no longer just distributed, but truly intelligent.
We support businesses in achieving truly intelligent and autonomous data architectures. Discover how our experts combine agentic AI and Data Mesh to industrialise your processes.
Designed by Zhamak Dehghani In 2019, the Data Mesh is based on four main principles: the decentralised data ownership by domain, the given as a product, the’self-service infrastructure and one federated governance balancing autonomy and compliance.
On paper, everything looks perfect: more agility, better data quality, and natural scalability. But implementation remains complex: distributed governance that's difficult to orchestrate, interoperability between domains, a lack of technical skills on the business side, and difficulty in discovering useful data.
Result: the promised autonomy can transform into new silos. The teams produce their Data Products but struggle to collaborate, generating integration debt and costly coordination.
One AI Agent is not a simple chatbot. It is a Autonomous assistant who perceives its environment, plans actions, interacts with various tools, and learns continuously.
Appliquéd to Data Mesh, these agents become:
They bring to the Data Mesh what it's still missing: a adaptive intelligence capable of handling complexity while preserving the autonomy of trades.
En Investment bank, An analyst requests in natural language a complex correlation between customer data and financial markets. Specialised agents automatically coordinate multiple business domains, resolve definition conflicts, apply compliance policies, and deliver the enriched report in minutes, compared to days previously. The semantic cartographer aligns terminologies, the controller verifies access rights, and the assistant generates the summary with full traceability.
In a Supply chain globally, agents proactively detect a risk of supply disruption, alert the relevant managers, propose optimised alternatives and automatically update ERP systems in real-time. Cross-domain orchestration is carried out according to explicit business rules, with the possibility of rollback and full audit.
En omnichannel retail, Client behaviour is continuously analysed by specialist agents, who automatically orchestrate personalised cross-domain campaigns while strictly adhering to GDPR consents and sensitive data policies.
These examples illustrate a major trend: agents are transforming the Data Mesh into a living, self-adaptive system, truly serving the business and its operational challenges.
By 2030, the most advanced companies will have a Intelligent domain meshing, each powered by its own AI agents.
The data will enrich one another, and decisions will be based on collective distributed intelligence.
The company will become data-aware : capable of detecting faint signals, learning continuously and adapting in real time.
In this model, the Augmented governance is no longer a post-hoc check, but a capability integrated into the architecture: policies, controls, and audits are natively embedded in the systems.
This evolution, however, presents several major challenges. The question of Transparency becomes essential: it must be possible to audit and understand the decisions taken by autonomous agents, even in complex contexts. Responsibility This also strengthens the legal and operational aspects: determining who is responsible in the event of an error, when multiple domains and agents are involved, becomes a key issue.
The transformation of professions also represents a significant organisational challenge. It is about supporting teams in this new agent-based environment without creating a technological divide or resistance to change. Finally, the’explainability automated decisions must remain a requirement, particularly in regulated sectors. These ethical and operational considerations are essential to ensure the responsible and sustainable adoption of agentic AI.
At JEMS, France's leading data industry player, we observe these changes up close from businesses and their real constraints. Our conviction is simple: the Data Mesh only exposes organisations to a complexity that is difficult to sustain in the long term, while the’Agentic AI remains ineffective without a solid organisational model and robust data foundations.
It is the intelligent combination of the two that paves the way for a new generation of businesses. autonomous, distributed and governed, capable of orchestrating their data continuously while maintaining control and consistency.
We support our clients in three complementary dimensions:
The merger of Data Mesh and of the’Agentic AI more than a technical evolution: it is a profound organisational transformation.
She promises more businesses agile, collaborative and intelligent, capable of transforming their data into a Sustainable competitive advantage.
But its success will depend above all on the ability to establish a Clear, auditable and humane governance, to hold business units accountable and to think of AI not as a substitute, but as a Collective facilitator.
At JEMS, we believe the future belongs to organisations that can transform their data into distributed, ethical and understandable intelligence.
Our approach aims to industrialise intelligence where it creates the most value, while preserving the balance between technology, governance, and people. It is this vision that we are already bringing to life alongside our clients.
