Enterprise architecture · Data and AI platforms
EA for Data & AI Platforms
Data and AI platform architecture that starts with decisions rather than licences: what the platform is for, what data earns a place, who owns it, which platform wins a clean selection, and how AI gets from pilot to production. One firm across architecture, data governance and AI governance, with no vendor alliances and every fee published.
The service at a glance
Definition
What is data and AI platform architecture?
Data and AI platform architecture is the decision layer above the technology: what the platform is for, what data earns a place on it, who owns that data, which platform wins a selection run clean of vendor influence, and how AI moves from pilot to production under real controls. Platforms fail in the gaps between architecture, data governance and AI governance, and this service exists to close all three sides of that gap.
Intelance provides it through services already priced on this site: twin baselines from £14,500 + VAT in ten working days, data ownership and standards built from £45,000 + VAT, architects and a senior data lead inside the build from £45,000 per programme and £22,000 per month, AI governance from £15,000 + VAT, and independent reviews from £8,500 + VAT. Every fee is agreed in writing before work starts.
A data platform strategy is a set of funded decisions rather than a reference architecture with your logo on it. Intelance holds no alliance with Databricks, Snowflake, Microsoft, AWS or Google, resells nothing and takes no commission, so the AI platform architecture is argued from your workloads and your bill. The wider practices are described under enterprise architecture consulting and data governance.
Use cases before sizing
The platform is scoped from what it must change rather than from a reference build.
Data with real owners
Critical data earns its place, with an owner who accepted the job.
A clean selection
The vendor chosen against your workloads, with no alliance in the room.
Pilot to production
AI leaves the sandbox under controls a regulator can read.
The decisions
Six decisions that decide whether the platform changes anything
They are made early in every platform programme, whether or not anyone on your side makes them. Each card shows the default you inherit if nobody does.
What the platform is for
The use cases that justify the spend, named and costed before anything is sized. A platform without a workload list is a subscription rather than a strategy.
Decided by defaultThe platform is for everything, which means it is for nothing anyone will fund twice.What data earns a place
The critical data domains that move first, chosen by value and quality, with the rest left where it is until it earns migration.
Decided by defaultEverything is ingested, the lake becomes a swamp, and trust dies in the first quarter.Who owns the data
Named owners for the domains the platform depends on, agreed before the build, because ownership is what keeps quality alive after go live.
Decided by defaultThe platform team owns everything technically and nothing actually.The platform chosen on the evidence
Databricks, Snowflake, Microsoft Fabric or the hyperscaler natives, argued from your workloads and your bill by people with nothing to resell.
Decided by defaultWhatever the incumbent integrator or the loudest vendor already sells.Pilot to production
The route from AI pilot to governed production: controls, monitoring and accountability designed before the tenth proof of concept.
Decided by defaultPilots multiply, production never arrives, and the board asks what the platform changed.The bill and the rules
Consumption modelled in full and the governance rules set before the meter starts, so cost and compliance are designed in rather than discovered.
Decided by defaultA consumption bill nobody predicted and controls retrofitted after the first incident.How it works
One firm across the whole platform journey
Each stage is an existing Intelance service with a published fee. Take one, or take the sequence.
The twin baselines
Systems and AI readiness scored from £18,000. The data reality profiled from £14,500. Together they tell you whether a platform is even the answer.
Ownership and standards
Named owners, standards and a measured quality baseline from £45,000 + VAT, because the platform inherits whatever governance exists, including none.
The platform chosen on the evidence
The selection run against your workloads and your bill, by a team with no alliance in the room and nothing to resell.
Architecture and data inside
Architects and a senior data lead inside the programme, holding the target and the data kept sound from design to cutover.
The production route
AI use governed onto the platform, and an independent review when the board wants proof of what changed.
Already bought the platform? The sequence still works: the baselines tell you what it should carry, and the foundations decide whether anyone will trust it.
Published fees
The four services platform programmes lean on hardest
Each one stands alone with its own published fee. Together they carry a platform from question mark to governed production.
Data Governance QuickScan
The data reality in ten working days: quality profiled, ownership mapped, and the gaps that would sink a platform found before it is sized.Best whenThe platform is still a plan and nobody has profiled the data. FeeFrom £14,500 + VATTen working days 02
Data Governance Foundations
Ownership, standards and a measured quality baseline built to stick: the governance the platform depends on, in place before the build leans on it.Best whenOwners, standards and quality do not yet exist in writing. FeeFrom £45,000 + VAT3 to 9 months 03
ProjectEA + Project DG
Senior architects and a senior data lead inside the platform programme: the target held, vendor designs challenged, the data rehearsed before cutover.Best whenThe build is real and the platform team marks its own homework. FeeFrom £45,000 + VATData lead from £22,000 / month 04
AI Governance QuickScan
AI use assessed and governed onto the platform: risk classified, controls set, and the pilot to production route designed.Best whenPilots are multiplying and production keeps not arriving. FeeFrom £15,000 + VATThe alternatives
Your options side by side
Three ways to architect a data and AI platform. The difference is what decides the design.
The platform vendor’s architects implement their own reference design
The consumption meter and the licence renewal
The reference architecture, whatever your workloads are
Teams that have already chosen and just need hands
A big firm data practice allied to Databricks, Microsoft or NVIDIA
You, and the alliance that feeds their pipeline
The practice’s accelerators and the partner roadmap
Enterprises that want scale and accept the alliance
A senior team across architecture, data and AI governance, from £14,500 + VAT to baseline
You, and nobody else: no alliances, no resale, no commission
Your use cases, your data quality and your bill
Organisations that want the platform to change something
Scope
What this includes and what it does not
You get
Twin baselines: the estate and AI readiness scored, the data profiled in full, in ten working days.
A workload list the platform is sized from, with the consumption bill modelled before signature.
A clean platform selection argued from evidence, on the record, with nothing resold.
Owners, standards and quality gates the platform can inherit, built to stick.
A pilot to production route for AI with controls a regulator, an auditor and a board can read.
This is not
An implementation. We build no pipelines and configure no clusters. The platform team and integrator build. We protect the decisions.
A reseller. No alliance with Databricks, Snowflake, Microsoft, AWS or Google, and no commission from any of them.
A data science bench. The models are yours. Our job is that they reach production under control.
A one-off report. Ownership and governance are built to run, and can be operated monthly through the Virtual CDO until your team is ready.
Where to start
Where clients usually start
Four situations we are usually called into. If none of them fit, book a call and we will point you at the right first step.
Why Intelance
The commitments we put in writing
Decisions before licences
The workload list, the data reality and the bill come before any platform is named. A platform without decisions is a subscription.
Vendor clean
No alliance with any platform vendor or integrator, no resale, no commission. The selection is argued from your workloads rather than our partnerships.
One firm across the gap
Platforms fail between architecture, data governance and AI governance. One senior team covers all three, so nothing falls between statements of work.
Published fixed fees
Every stage is an existing service with its fee published on this site, from £14,500 + VAT, agreed in writing before work starts.
Governance built to outlast us
Owners who accepted the job, standards teams use and a baseline you can measure: run by your people, or operated monthly through the Virtual CDO and AI Governance Office until they are ready.
Who you will work with
Every engagement is led by a senior partner. Nothing is offshored and nothing is handed to a junior bench.
Case studies
All case studiesReady to talk?
Tell us where the platform stands
Thirty minutes with a senior architect on where things are: being pitched, being built, or already bought and strangely quiet. If a vendor deck is on the table, say so. The selection work is designed to land before signature.
FAQs
Questions about EA for data and AI platforms
What is data and AI platform architecture?
The decision layer above the technology: what the platform is for, what data earns a place on it, who owns that data, which platform wins a clean selection, and how AI moves from pilot to production under real controls. It sits across enterprise architecture, data governance and AI governance, because platforms fail in the gaps between the three.
How much does data and AI platform architecture cost?
With Intelance it is priced through the services that carry it: the Data Governance QuickScan from £14,500 + VAT and the EA QuickScan from £18,000, both in ten working days. Data Governance Foundations from £45,000 + VAT. Architects and a senior data lead inside the build from £45,000 per programme and £22,000 per month. The AI Governance QuickScan from £15,000 + VAT. Independent reviews from £8,500 + VAT. Every fee is published and agreed in writing before work starts.
Do we need a data platform?
Sometimes the honest answer is no, or not yet. The twin baselines exist to answer that question with evidence: if the use cases do not justify the spend, or the data quality would poison the platform on arrival, you find out in ten working days for a five figure fee rather than after a seven figure build.
Which platform do you recommend: Databricks, Snowflake or Fabric?
None of them by default, which is the point. Intelance holds no alliance with any of the platform vendors, so the selection is argued from your workloads, your data gravity, your team and your bill. The right answer differs by organisation, and a firm with a partnership to feed cannot tell you that.
What is a data platform strategy?
A set of funded decisions: the use cases the platform must serve, the data domains that move first, the owners who keep them honest, the platform that won a clean selection, the route from AI pilot to production, and the run cost modelled before signature. A reference architecture with your logo on it is not a data platform strategy.
Our last platform changed nothing. Why would this one?
Because the last one probably inherited the six defaults: a platform for everything, all data ingested, no real owners, a vendor chosen by alliance, pilots without a production route, and a bill nobody modelled. The fix is not a better platform. It is making the six decisions deliberately, which is what this service does.
How does AI get from pilot to production?
Through a governed route designed before the pilots multiply: use cases classified by risk, controls and monitoring set to match, accountability named, and the platform configured so the compliant path is the easy path. This is where enterprise architecture for AI earns its keep, and it is the job of the AI Governance QuickScan and the AI Governance Office.
Do you build the platform?
No. We build no pipelines, configure no clusters and write no models, which is what keeps the advice clean. Your team and your integrator build. We make sure what they build is what you chose, and that it is governed well enough to be trusted.
Who runs the governance after go live?
Your people, by design: owners who accepted the job, standards they use and a quality baseline they can measure. Where the team is not ready, the Virtual CDO runs data governance monthly from £7,500 + VAT and the AI Governance Office runs AI governance retained, with the exit designed in.