1 of 20

Data Mesh – Fundamental Pillar for AI-readiness

When AI Meets Data Reality: How to Manage and Govern Data to Enable Swisscom for AI-Readiness

2 of 20

How it all started

The beginning of our Data Mesh journey

Mirela Navodaru | Data Mesh Live 2026 | C2 General

2

3 of 20

Mirela Navodaru | Data Mesh Live 2026 | C2 General

3

with People for People

with

People & AI

for

People & AI

4 of 20

AI-readinessMeasures an Organization’s preparation to adopt, deploy, and scale AI effectively��

Mirela Navodaru | Data Mesh Live 2026 | C2 General

4

Main Success Factors

  • Business strategy
  • AI Technology infrastructure
  • Data Product Foundation
  • People & Culture
  • Governance

AI-readiness Levels

Data needs to be:

findable and described

trusted and contracted

semantically navigable

5 of 20

��In the AI era�Data Mesh is here to stay

How Data Mesh pillars contribute to the AI-readiness

Mirela Navodaru | Data Mesh Live 2026 | C2 General

5

6 of 20

Organization –Data & AI is everyone’s business

Mirela Navodaru | Data Mesh Live 2026 | C2 General

6

SPOKES

HUB

Data, Analytics

& AI

Value Spokes

(Business Units)

Master Data

Producers

(Operational Source Systems)

70%

of AI Investments

should go into

People and processes

Source: McKinsey

7 of 20

Mirela Navodaru | Data Mesh Live 2026 | C2 General

7

Federated Governance – Shared Responsibility

CENTRAL GOVERNANCE

(What & Why)

  • Architecture principles
  • Naming conventions & standards
  • Compliance rules & policies
  • Data product lifecycle processes

DECENTRALIZED DATA TEAMS

(How)

  • Implement governance locally
  • Define data quality
  • Closest to data, consumers and risks
  • Push back when rules meet reality

8 of 20

Mirela Navodaru | Data Mesh Live 2026 | C2 General

8

Swisscom Data Model (SDM) - a common language for data exchange

Central (Data Management and Governance): owns, develops and maintains the conceptual & logical model across domains.

Data Teams: use it as the reference for every data product (logical and physical data models) — and extend it where their business needs more.

Aligned with TM Forum standards

Shared semantics and consistent relationships is what enables AI to reason across domains

9 of 20

Mirela Navodaru | Data Mesh Live 2026 | C2 General

9

Principles

Components

Data Product - the unit of trust in the mesh

Data

Metadata

Code

Policies

A named owner

One team, one accountable Data Product Owner.

Discoverable

Registered in the data product catalog.

Value-driven

An explicit value proposition: what problem it solves, who consumes it, and why.

Built on the SDM

Reusable and interoperable through the common language.

Shared via Contract

Explicit claims between producer and consumer.

Trustworthy

Security and compliance rules built in from the start.

Context

10 of 20

Mirela Navodaru | Data Mesh Live 2026 | C2 General

10

Data Products – types, usage, conformity

SOURCE SYSTEMS

Source-aligned

Closest to the operational systems — owned where data is born.

Aggregated

Combining entities into broader, reusable views.

Consumer-aligned

Data gains business meaning at every step.

SDM CONFORMITY LEVEL FOR DATA PRODUCTS – BASED ON PURPOSE

GOLD core SDM conformance

SILVER non-analytical purposes

BRONZE analytical purposes

CONSUMER

Analytical purposes

Non - analytical purposes

Consumer-aligned

Shaped for a specific consumption purpose.

11 of 20

Mirela Navodaru | Data Mesh Live 2026 | C2 General

11

Architecture & Technology - two platforms: one for data, one for meaning

ONE DATA PLATFORM (ODP)

  • Hybrid data platform (on-prem & cloud)
  • Self-service technical enablers
  • Data flows and Data Assets (tables, views, Kafka topics)
  • Patterns: ingestion, transformation, usage

Domain- and use-case-agnostic: the same road for every team.

COMPASS

METADATA & CONTEXT MGMT. PLATFORM

Ownership · Schema

findable & described — Level 1

Quality and Compliance rules

trusted & contracted — Level 2

Semantics · Taxonomy · Ontologies

semantically navigable — Level 3

ODPS & ODCS to govern at scale and provide e2e Data Observability & Mesh Health

12 of 20

Changing the airplane engine�while flying

Transforming to Data Mesh

while going from on-premise to a hybrid Data platform

while enabling an AI-first company

Mirela Navodaru | Data Mesh Live 2026 | C2 General

12

13 of 20

Opportunities and trade-offs in challenging times�

Mirela Navodaru | Data Mesh Live 2026 | C2 General

13

Opportunities

Enablers:

  • Governance as Code
  • Metadata as Code
  • Agentic Code

Trade–offs accepted during Migration

  • Automation of Data Product Creation
  • Automation of Self-assessment for Compliance
  • Talk to Data

Business continuity beats purity

  • Exceptions as part of the process
  • Tracked and documented
  • Plan and Roadmap with deadline for sanitization

14 of 20

Lessons learned from our Journey

What worked and what needed improving

Mirela Navodaru | Data Mesh Live 2026 | C2 General

14

15 of 20

Mirela Navodaru | Data Mesh Live 2026 | C2 General

15

Top Management commitment

A concrete AI-first ambition settles every priority fight.

Governance

Maintain discipline and interoperability in the decentralization

Stay flexible and adapt

Reality check, identify opportunity, stay strategic

Operating model First

Early clarity on who does what, and why, fill the gaps in skills & culture

Quicker in making decisions

Try, learn, iterate: a good-enough decision today beats a perfect one next quarter.

Success

Improving

Make technology easy to use

Self-service and automation for the technical enablers

16 of 20

Things to take home

Key Takeaways and Conclusion

Mirela Navodaru | Data Mesh Live 2026 | C2 General

16

17 of 20

Key Takeaways

Mirela Navodaru | Data Mesh Live 2026 | C2 General

17

One

Data Mesh is the foundation for AI-readiness

Data Product Ecosystem with clear ownership and governance to produce and exchange data for humans and AI.

Two

There is no such thing as a green field

Transform while operating. Brownfield is not the obstacle to the strategy, find the right balance of governance and flexibility

Three

Opportunities in challenging times

Don’t just find better ways to reach your objectives , but let the shift set new and bolder ones.

18 of 20

Mirela Navodaru | Data Mesh Live 2026 | C2 General

18

Conclusion

In the Agentic era, People must remain in the Lead of Design and Architecture for the AI-readiness foundation to avoid

Garbage – In,

Danger - Out

19 of 20

Mirela Navodaru | Data Mesh Live 2026 | C2 General

19

By making Data and AI everyone's business,

We contribute to a safer AI future.

20 of 20

Mirela Navodaru�Enterprise and Solutions Data Architect

LinkedIn

20

Thank you

for your attention! �