Services · Implementation · Step 3 of 3

From prototype to production in 8–12 weeks.

Fabric Accelerator is NuData's co-production program: we design the architecture and supervise the build while your team executes it, moving through phases with clear deliverables. By the end, your team is the technical owner of everything built, with no vendor lock-in and no black boxes.

8–12 weeks · fixed scopeUSD 15,000–30,000 · no variable costsPhased co-production · your team learns by doingFull technical ownership

The industry's problem

Pilot Purgatory.

Most data projects don't fail: they get stuck. They work in the sandbox but never reach real production.

80–90%

of AI projects never reach production

Not for lack of talent — for lack of a solid data foundation and governance from the first sprint.

Microsoft Build 2026

46%

of AI proof-of-concepts are scrapped before reaching production

The team keeps working, the results never arrive.

S&P Global Market Intelligence — Enterprise AI Survey 2025

95%

of those projects share the same root cause

Architecture without governance by design, an ambiguous semantic model and pipelines without data quality monitoring.

IDC Worldwide Data Management Survey 2025

The Accelerator attacks all three causes in a single program. Build it right from the start, with the team that will operate it, under the direct supervision of a Microsoft Certified Trainer.

The situation assessment

Sound familiar?

The prototype never reaches production

It works in the sandbox, but promoting it surfaces performance problems, inconsistent data or security rejections.

Rework that never ends

Every sprint solves one problem and creates two. The architecture doesn't scale, pipelines break when the schema changes.

Copilot that hallucinates

They connected Copilot to the existing model and the answers are wrong. Users don't trust it, risk blocked the rollout.

Vendor lock-in through technical dependency

The previous consultancy delivered something nobody on the team understands. Every change is a new purchase order.

At a glance

Duration8–12 weeks depending on scope and team size
FormatPhased co-production · NuData designs, your team builds
Artifacts5 technical deliverables with ownership transfer
Who attendsHead of Data · Architect · Engineering team
PrerequisitesMicrosoft Fabric tenant · technical team available
Next stepYour team operates autonomously
Led byNuric Ugarte · MCT · Microsoft author · Fabric consultant

The methodology

Guided co-production, in five phases.

We don't deliver black boxes. The methodology is designed to transfer technical judgment, not just code. The project moves through phases, each with its own deliverable:

01 · Foundations

We survey your environment and design the adapted Medallion architecture, validated with your team, with every decision documented along with its why.

02 · Implementation

Your team builds OneLake, pipelines and governance under direct supervision: weekly working sessions and a Q&A channel between sessions.

03 · Validation

We check what was implemented against the design: load, performance and consumption tests, plus governance (roles, RLS, classification).

04 · Upskilling

Intensive training on your real environment (not toy data), with a final team assessment.

05 · Handoff

Delivery with complete documentation, an executive report and a 6-month roadmap. Your team is left operating on its own.

Who does what:

NuData brings the architecture design, the governance and security decisions, the technical direction and the quality validation.

Your team builds on its own infrastructure, guided through every phase.

The outcome at the end:

Your team maintains and extends everything that was built

Every architecture decision is documented and justified

The risk team can audit full lineage and access

You don't depend on NuData to operate

No black boxes nobody understands

No vendor lock-in or recurring purchase orders

What gets built

Five technical artifacts. All owned by your team.

01

Certified Medallion architecture

A lakehouse with Bronze/Silver/Gold layers on OneLake, designed to scale and documented so anyone on the team can operate it.

“For the first time we have an architecture the team understands and can maintain on its own.”
— Head of Data
02

Data pipelines in production

Ingestion, transformation and load with quality monitoring, alerts on schema failures, refresh SLAs and documented lineage.

“I know exactly where every number on the dashboard comes from and when it was updated.”
— Data architect
03

Semantic model validated for AI

A coherent ontology, unambiguous measures and correct RLS. Validated as a reliable foundation for Copilot and Microsoft Fabric agents.

“The risk team approved connecting Copilot because they can see exactly where the answers come from.”
— CIO
04

Governance and security framework

Access in OneLake, Purview for lineage and classification, documented roles and onboarding for new team members.

“The next security audit will be the first one that doesn't scare us.”
— Head of Technology
05

Technical knowledge transfer

Your team builds under guidance; by the end, every pipeline and every decision has an internal owner who understands it.

“My team learned more in 8 weeks than in the previous 18 months.”
— Head of Data

How it works

From the first call to technical ownership.

01

Scoping call — 30 min

If you're coming from the Clarity or the Blueprint, we bring the findings and the already-validated design as the starting point.

02

Fixed proposal — 48 h

Scope, week-by-week timeline and total cost. No variable costs.

03

Foundations phase — Weeks 1-2

Medallion architecture, sources, workspaces and governance plan, reviewed with your team before writing any code.

04

Guided implementation — Weeks 3 to 10

NuData designs and supervises; your team executes. Weekly progress and quality reviews.

05

Validation, Upskilling and handoff — Weeks 11-12

Load tests, quality validation, security approval, training on your environment and delivery of the knowledge repository. You walk away with full technical ownership.

Investment

Two tiers based on your starting point.

Fixed price. No variable costs. No surprises at the end.

8–10 weeks

Team Enablement

USD 15,000–20,000

Who it's for: Teams with Fabric live that need to go from prototype to production with a solid foundation and no lock-in.

  • Certified Medallion architecture
  • Pipelines in production
  • Semantic model validated for AI
  • Complete governance framework
  • Technical knowledge transfer
Book Team Enablement
Most popular

10–12 weeks

Comprehensive

USD 20,000–30,000

Who it's for: Organizations migrating from Power BI, Synapse or legacy platforms, with technical debt remediation.

  • Everything in Team Enablement
  • Exhaustive migration review
  • Platform comparison
  • Executive justification
  • Extended post-delivery support
Request a proposal

If you're coming from the Fabric Clarity Check (50%) or the Fabric Blueprint (50%), that investment is credited within 30 days.

Your team can run Fabric in production, and own every technical decision.

30 minutes with a NuData architect to design the Accelerator scope specific to your context.

Frequently asked questions

Does the internal team need prior experience with Microsoft Fabric?

It's not required: the co-production is designed so they learn by doing, guided by a Microsoft Certified Trainer. They do need experience with SQL, data and basic cloud concepts.

How is the work split?

NuData designs, supervises and validates; your team builds on its own infrastructure, guided through every phase. They learn by doing and understand every pipeline by the time we leave.

How is this different from a traditional consultancy?

We don't deliver black boxes: your team builds and ends up as the technical owner, with no recurring purchase orders. When we leave, your team can maintain, extend and audit everything that was built.

Does it include preparation for Copilot and AI agents?

Yes: the semantic model is built and validated for Copilot and Fabric Data Agents, with semantic coherence checks, correct RLS and ontology documentation your risk team can audit.