Intelligent Business Applications Lead
Turn governed enterprise data into intelligent applications that help people decide, act and automate.
Analytics remains essential. But insight does not always need to be the end of the workflow.
The opportunity
And your application developers no longer need to start by rebuilding the data foundation. That is the whole opportunity, and it is mostly an organisational one.
Most large organisations have spent years making enterprise data trustworthy. Domains have owners. Definitions are agreed. Reporting is governed. That work was necessary and it was expensive.
Meanwhile the operational work that depends on that data still happens somewhere else — in a spreadsheet, in an inbox, in a meeting held because nobody could see the whole picture in one place.
The gap between those two facts is where intelligent business applications live.
Disciplines that have to converge
None of these is new. The value comes from combining them around a single decision, rather than organising them into separate queues.
Wherever you are with the platform
Already established
You have already paid for the hard part. The question is what else it should be carrying.
Being implemented now
The best moment to decide that this platform is not only going to produce reports.
Being evaluated
Judge the platform on everything it could carry, not only on the dashboards it replaces.
You have already made the expensive investment. The question is whether it stops at reporting.
The same foundation, two destinations
Both paths start from the data you already govern. One ends when someone understands the situation. The other continues until something in the business has actually changed.
Both paths begin at Governed enterprise data. The analytics path has 3 stages and ends at Insight. The application path has 6 stages and continues past that point through Business context, AI, Application, Decision, Action, Feedback — it extends the analytics path rather than replacing it.
Shared starting point
Governed enterprise data
The investment you have already made, or are making now.
01Semantic model
One agreed definition of the business measures.
02Analytics
Reporting, exploration, monitoring.
03Insight
Someone now understands what happened.
This is where most data platforms currently stop. That is not a failure — it is simply not the whole return on the investment.
01Business context
The relationships and rules your people already use.
02AI
The situation summarised, the likely cause proposed, the impact estimated.
03Application
A person sees the case, the evidence and the options.
04Decision
Someone chooses, and the choice is recorded.
05Action
A work order, a re-plan, an escalation. Something actually changes.
06Feedback
The outcome becomes data the next decision can learn from.
Path B can run alongside Power BI, after it, or on governed data where no reporting practice exists yet. It is an additional way to extract value from the same foundation.
So which decisions in your business are worth moving from a report into an application?
Start with the outcome
Pick a challenge and an industry. The recommendations below come from a published opportunity model, not from a language model — the same answers, every time, for the same inputs.
Most relevant in Manufacturing
Manufacturing
One place where a production deviation is detected, explained, costed and resolved — instead of three systems and a phone call.
Decision
Given this deviation, what is most likely causing it and what is the least costly intervention right now?
Manufacturing
Rank maintenance work by what it protects — production commitments, safety and cost — rather than by when the ticket was raised.
Decision
Of everything we could do in the next maintenance window, what should we actually do?
Manufacturing
Bring process conditions, material batches and inspection results together so root cause analysis takes hours rather than weeks.
Decision
What process or material conditions are associated with this quality deviation?
Data domains involved
Likely first pilot
Production Exception Manager
That is the shape of the opportunity. This is what it looks like when someone actually uses it.
Industry relevance
Give them a governed path to the solution. Choose an industry to see where the value usually sits — then use the application, because reading about one proves nothing.
Your plant reports OEE every morning. The decision that protected this morning's OEE was made at 03:20 by a shift supervisor with a phone and a spreadsheet.
Manufacturing already measures almost everything. The gap is rarely visibility — it is the distance between noticing a deviation and doing something about it while it still matters.
All Manufacturing opportunitiesData domains
Where the data comes from
Flagship experience
One place where a production deviation is detected, explained, costed and resolved — instead of three systems and a phone call.
This is the difference. A report would show you the deviation. Here you open it, read the reasoning, weigh the options and commit an action that leaves the screen.
Other opportunities in Manufacturing
Bring process conditions, material batches and inspection results together so root cause analysis takes hours rather than weeks.
Rank maintenance work by what it protects — production commitments, safety and cost — rather than by when the ticket was raised.
Show which inbound disruptions actually threaten a customer commitment, and what the realistic alternatives cost.
Let engineers ask questions of specifications, change history and past deviations in language, with citations back to the source document.
One person, one decision, a few minutes saved. Multiply that by the number of people who make it.
Business value
An indicative opportunity, calculated from your own assumptions. Not a saving, not a guarantee, and not a business case — a starting point for deciding whether this is worth a proper one.
Hours returned to the business
9,200
hours / year
≈ 5.8 full-time equivalents
Potential annual capacity value
€690,000
indicative
from 22,080 events a year
The calculation
If that is worth having, the next fair question is how it is actually built and operated.
How it actually works
You do not need to understand every layer. You do need to know that the application is built on the same governed data your reporting already trusts.
The architecture is a stack of 10 layers, from the bottom up: The systems you already run on, then Getting the data out reliably, then One governed copy of the truth, then Data owned by the people who understand it, then Understanding what happened, then Teaching the platform your business, then Judgement at the speed of the process, then Where people actually work, then Something actually changes, then Doing this safely, repeatedly. Selecting a layer below describes it in full.
The layers
Layer 8 of 10
The place a person sees the situation, the evidence and the options — and makes the call. This is the layer most data platforms never reach.
Data flows
The general pattern. Data the business already produces becomes a governed product, gains business meaning, and ends up in an application where someone decides and acts.
Enterprise data to application, end to end: Systems you already run on, which feeds Reliable delivery, which feeds One governed copy, which feeds Owned by the business, which feeds Agreed business entities, which feeds Business meaning, which feeds Explanation and prediction, which feeds Where people decide, which feeds Something changes. The flow ends in a change to an operational system, not in a report.
Switch to Technology view in the header to see the named capabilities behind each layer, their maturity, and the two hosting patterns worth considering.
One application proves the case. It is not the point.
The operating model
The people closest to the problem can now prototype far more of the solution themselves. Handled well that is governed innovation. Handled badly it is Shadow IT with better tooling.
Business builder
Understands
Contributes
Industrialisation
Understands
Contributes
The paved road
Understands
Contributes
Accelerate innovation and reduce Shadow IT at the same time. Switch to Technology view for what the platform team actually has to provide.
From one app to an App Factory
One high-value application, in production, changing a decision that people can point to.
You know you are here when
Before we talk
Seven questions, scored transparently in your browser. Nothing is uploaded, stored or tracked.
Engagement model
A practical sequence rather than a programme. Each stage produces something you can act on, and each one is optional if you already have what it produces.
Find where an application would change a decision that actually matters, and prove the case is real.
Typical work
Deliverables
Durations are indicative and depend on scope, data access and how many decisions the first application has to reach. No pricing is published here — that conversation belongs with the scope.
If your organisation has — or is building — a serious data platform, there is probably more value in it than another dashboard.
Okko Oulasvirta
Intelligent Business Applications Lead
For the last decade I have worked inside large European industrial and data-intensive enterprises where cloud, data, software engineering, DevOps and business operations meet.
I help organisations connect the data platform, AI capability and software engineering model required to turn this opportunity into an operating capability.
For larger engagements I bring together a trusted delivery team across Data & AI, application development, UX, integration, DevOps and platform engineering.
Illustrative demonstration. Northstar Industrial, its plants and all data shown are fictional. The generated assessment is deterministic and requires no AI service.