dLocal

dMesh has lowered the bar for using data. Traditional self-service analytics expected people to know SQL, understand table schemas and find their way around complex data models. dMesh flipped that.

Through Databricks Genie, integrated with conversational AI tools like Glean and Claude, anyone at dLocal can ask a question in plain English and get a trusted, governed answer in seconds. A Technical Account Manager can ask "Which Brazil merchants haven't adopted Pix?" and surface cross-sell opportunities straight away. A product manager can validate A/B test results without filing a data request. A finance analyst can pull reconciliation data without waiting on the central team.

The conversational layer sits on a semantic layer and governed metric views, so when someone asks about "conversion rate" or "TPV", they get the same certified definition as everyone else in the organisation.

The result is that the time from question to trusted answer is measured in minutes, not days or weeks.

dMesh makes producing data easier too, through AI-assisted tooling and engineering best practice. Domain teams, the people who actually understand the data, can now produce certified data products using:

• AI Skills that build the first version of a new data product while cross-checking it against existing ones.
• A Data Product Factory with templated CLI tooling, so creating a new data product follows a guided, repeatable process instead of a bespoke engineering project.

A domain team member starts a new data product with a single CLI command and fills in a YAML data contract defining its schema, input sources, SLAs, data classification and owners. The platform handles the rest: governance, tagging, access policies and deployment.

AI lowers the barrier further by helping with transformation logic, quality rule generation and metadata enrichment. Domain teams spend their time on business logic rather than infrastructure plumbing.

dMesh uses a federated governance model that moves accountability for data to where the expertise actually lives, in the business domains. Rather than funnelling every decision through a central committee, governance is computational and built in by design.

Three roles make this work for every data product:

• Data Owner: accountable for the data product overall, making sure it solves real business problems and stays high quality. Typically a Head or VP within the domain.
• Business Data Steward: responsible for the product's definitions, oversight of its delivery pipeline and where its data comes from, keeping its meaning aligned with the business.
• Technical Data Steward: responsible for implementation, pipelines, contracts and incident response, keeping the product running and evolving.

Governance is part of the platform, so it never becomes a gate:

• Unity Catalog provides central access control, lineage and audit across all data products.
• Every asset is automatically tagged for cost and usage attribution by domain.
• PII and PCI classification, hashing, masking and access policies are applied automatically through the data contract.
• Security controls are enforced at platform level, so compliance is continuous and auditable and does not slow teams down.

A Data Stewards Working Group meets monthly to resolve cross-domain issues, and a Data Governance Council gives quarterly sign-off and settles conflicts. Day-to-day governance happens automatically through the platform, with no manual gatekeeping.

The programme runs as four progressive modules, delivered as bite-sized content on a microlearning platform and building from foundational awareness to applied analytical thinking:

• Analytics Savvy: a light-touch introduction to data behaviours for the whole organisation, building awareness without overwhelming non-technical staff.
• Data Accountable: deeper content on data ownership, governance responsibilities and the feedback loop between data consumers and producers.
• Analytics Powered: practical skills for using data products and AI-powered analytics in day-to-day decisions.
• Data Curious: advanced questioning techniques, recognising bias and creative approaches to exploring data.

The curriculum focuses on lasting behavioural skills rather than tool-specific training. Tools change. The ability to ask a sharp question, challenge an assumption and think critically about what the data says does not.

dRiven launched with a pilot cohort of around 100 data stewards, domain champions and cultural ambassadors, who tested each module, gave structured feedback and helped refine the content before the company-wide rollout. Starting with a pilot kept the programme grounded in dLocal's real context rather than generic data literacy theory.

The full launch ran across several channels: CEO-endorsed communications, office-based "Launch Labs" with live demos, a dedicated community channel for peer exchange, and video messages from the CFO, CTO, COO and CRO.

Rather than treating data literacy as a bottom-up training exercise, dRiven enlisted senior leaders to role-model data-informed behaviour, endorse the initiative publicly and connect it to their teams' objectives.

Each C-suite leader recorded a personal message tying data capability to concrete outcomes for their function: better forecasting and capital allocation for the CFO, faster diagnosis and prioritisation for the CTO, earlier issue detection and consistent execution for the COO, and sharper commercial decisions and growth for the CRO.

dRiven tracks engagement depth rather than course completions: how many people are actively asking questions in the platform, sharing insights with peers and applying data habits in their work.

On launch day alone, 106 new users accessed the platform, 96 ran their first query, 40 joined the community channel and 9 shared their results publicly, none of it mandated.

The communications plan moves deliberately from launch awareness to practical relevance to community-led adoption, with regular checkpoints to decide whether more intervention is needed. The goal is a lasting organisational habit as dLocal grows, rather than a one-off training event.

The data mesh vision is often criticised as theoretical or impractical. dMesh shows it can be delivered at scale with the right platform, engineering discipline and partner. The key lessons from dLocal's journey:

• Self-service consumption through AI is real. It is a production capability serving hundreds of users daily, well past the demo stage. Natural language interfaces over governed semantic layers remove the SQL barrier without sacrificing trust.
• Production can be democratised. Give domain teams templated, code-first tooling with governance built in, and they can produce certified data products without deep platform engineering skills.
• Engineering discipline is non-negotiable. Data contracts as code, CI/CD for data products and automated quality enforcement are what separate a working data mesh from a collection of ungoverned tables.
• Federated governance works when it is computational. Ownership, stewardship and compliance belong in the platform by design rather than in committees and policy documents.
• Adoption follows value. dMesh's user growth came from making data genuinely easier to find, trust and use, without mandating adoption.
• Cultural change is essential. Technology alone does not change behaviour. A deliberate programme that builds curiosity, critical thinking and data confidence, endorsed by the CEO and C-suite, turns a platform launch into an organisational habit.