5%
of the enterprise generative AI pilots MIT studied produced a measurable change in revenue or cost. The other 95 per cent did not.
Source: MIT NANDA and MIT Media Lab, The GenAI Divide: State of AI in Business 2025.
Operance AI
We are an AI consultancy and venture studio for leadership teams, and we turn AI spending into results that show in your numbers. We start from one decision you want made better and work across people, processes and platforms to improve it. Everything we set up is yours, in writing, and you stay free of any one vendor.
Your board is asking what AI has delivered.
The licences are paid, a pilot or two is running, and people like the tools. Yet when the quarter closes, nobody can point to a number that moved.
About 5 per cent of the pilots MIT studied did produce measurable value. The companies behind them have an answer when the board asks. Getting your company into that group is our job.
Everything we do for you starts with one decision to improve, one owner and a way to measure the result from day one. If we cannot find a decision worth improving, we say so.
Published research
40%+
of agentic AI projects, where AI carries out tasks on its own, will be cancelled by the end of 2027, Gartner expects. The reasons it gives are escalating costs, unclear business value and inadequate risk controls.
MIT's researchers found that most generative AI systems “do not retain feedback, adapt to context, or improve over time.”
What changes when we set up AI in your business.
Every company can buy the leading AI models at the same price, on the same day. Used as they come, they give much the same answers.
We bring your data together, train a model on your own material and put AI agents to work in your systems. After each cycle we measure the result and feed it back in.
Diagram of AI in a company, before and after we set it up
Before, most companies use the same AI model as it comes, with their data spread over systems and a pilot or two. After, four parts are set up and connected to the systems your people already use, and the results go back into them. The four parts are listed after the diagram, each with its before and after.
Before, What most companies have now
-
Your data
Spread over systems, stitched by hand
-
The same model
Used as it comes
-
At work
A pilot or two
-
Output
- Much the same answers
- No number that moved
-
After each cycle
Nothing fed back
After, All four parts, connected to the systems you already use
The systems your people already use
Your ERP, your CRM, your warehouse system
-
Your data
One set of numbers everyone trusts
-
The same model
Trained on your own material
-
At work
AI agents in the systems your people use
-
Output
- Answers people can check before they act
- Routine work done, every action logged
- Each improvement measured before and after
-
After each cycle
Testing and monitoring
Before and after, part by part
- 1Data management, Can start here. Before: Meetings argue about whose number is right. After: One set of numbers everyone trusts
- 2Model training, Can start here. Before: The model everyone has, used as it comes. After: Trained on your material, accuracy tested
- 3AI agents, Can start here. Before: A pilot or two, and tools few people use. After: Routine work done in your own systems
- 4Testing and monitoring, Can start here. Before: Nobody can point to a number that moved. After: Each improvement measured before and after
Where to start
We set up all of it, connected to the systems you already use, or just the part you need.
- All of it. Nothing new to log in to, so nothing gets skipped.
- Data management. One set of numbers everyone trusts, so meetings move on from whose number is right to what to do next.
- Model training. A model trained on your material, with its accuracy tested and on record.
- AI agents. Routine work done without someone driving it, and every action logged, so an audit is a quick search.
- Testing and monitoring. Each improvement arrives with its before and after numbers on record.
We set up all four parts. Nothing new to log in to, so nothing gets skipped.
One cycle, repeated until the results show.
Before AI takes on any work, we agree with you which decision should get better and how the result will be measured. Then AI carries out the agreed task, we measure what changed, and each cycle starts from what the last one taught. We call it the loop.
- Intent: We agree which decision should get better.
- Execute: AI carries out the agreed task.
- Measure: We measure what changed.
- Learn: The next cycle starts from what this one taught.
And round again What we learn sets the next intent, and a new cycle begins.
AI pays when people, processes and platforms move together.
The AI Transformation Triangle is our own framework. People decide whether AI gets used, processes decide whether it delivers value and platforms decide whether it scales. Many organisations are strong on one side and thin on the other two. Some have good tools that few people use. Others have keen teams and no process to put the tools in.
Results come when the three work together. The work gets done a better way and new tools reach the people they were meant for. Your systems carry the work from start to finish. Together these form your operating model, and that is what we redesign with you.
Hover over or tap a side or a layer of the triangle to see what it covers.
- People
- Everything that decides whether AI gets used.
- Processes
- Everything that decides whether AI delivers value.
- Platforms
- Everything that decides whether AI scales.
- Operating layer people × processes
- How the work gets done once AI is part of it.
- Adoption layer people × platforms
- Whether new tools reach the people they were meant for.
- Orchestration layer processes × platforms
- Whether your systems carry the work end to end, without a person stitching them together.
People
Everything that decides whether AI gets used.
Processes
Everything that decides whether AI delivers value.
Platforms
Everything that decides whether AI scales.
Plan it, put it to work, keep improving it.
Any of the three is a place to begin. The strategy tells you which work is worth paying for before you spend more.
Transformation
We redesign how your organisation works with AI.
We set the strategy with your leadership team and choose the decisions where AI makes the biggest difference to your results. These can be as everyday as which quotes to chase, what to keep in stock or which customers to call this week. Then we redesign how that work is done.
What you get
A strategy you can act on, tied to the decisions it is meant to improve. Each decision we examine gets a clear go or no-go.
Implementation
We put AI to work in your business.
We bring AI into the systems your people already use. We also set up the data and the ways of working that make it part of the daily routine.
What you get
AI doing real work in your business, and everything we set up is yours from day one, in writing.
Optimisation
We close the loop, so each cycle improves on the one before.
We measure what changed and feed each result back in. Every lost quote is recorded with its reason, so the next cycle can learn from it.
What you get
An advantage that grows with every cycle and is hard for a competitor to copy. Each cycle ends with its before and after results on record.
Twelve services, and you pay for what your problem needs.
We use a different mix each time, chosen for the problem at hand.
Strategy
Which decisions are worth improving with AI, what moving each one is worth, and what it would take. Done before any money goes into systems.
A clear go or no-go on each opportunity, with numbers your finance team can check.
Education
Managers learn to read what the AI tells them. The people whose work changes are brought along from the start, so the new way of working lasts.
Teams that understand what changed and use it in their daily work.
Data management
Where your numbers live, how they are kept accurate, and a starting measurement taken before anything changes.
One set of numbers everyone trusts, so meetings move on from whose number is right to what to do next.
Models
For when an off-the-shelf model is not enough. We train one on your own material, to sort your documents, read your images or do one task better than a general model.
A model trained on your material, with its accuracy tested and on record.
Generative AI
Drafting, summarising and answering questions from your own documents and data, with every answer traceable to its source.
Answers people can check before they act, and regular tests that show how often they are right.
AI agents
AI that carries out a routine task on its own. Each agent has one named job, written limits on what it may decide, and a person who handles the exceptions.
Routine work done without someone driving it, and every action logged, so an audit is a quick search.
Applications
The screen your people open every day, designed around their working day so it is still in use on a busy Tuesday afternoon.
A tool people choose to use, owned by your organisation.
Integrations
AI connected to the tools your people already have open: your ERP, your CRM, your warehouse system.
Nothing new to log in to, so nothing gets skipped.
Monitoring
Every change is tested before your people see it. Once it is live we watch it, so a drop in quality shows in your own reports.
Each improvement arrives with its before and after numbers on record.
Responsible AI
What the AI may decide on its own, what goes to a person, and how you would show both to a regulator. European rules, the EU AI Act, are coming into force in stages, and designing for them now costs less than adding them later.
A written record of what it decides alone, where a person signs off and how it was tested, ready before anyone asks.
Security
AI that acts on its own can be tricked or misused. We test it the way an attacker would, from hidden instructions to data leaving by the wrong route, and close the gaps we find.
Every test on record, and the gaps we found closed before launch.
Managed AI
Once it is live, we can run it for you: questions answered, models replaced when a supplier retires one, quality watched and costs explained.
When a supplier retires a model, the switch is already planned. You choose in writing whether we run it or your team does.
Operance AI Studio is where we build.
Operance AI Studio is our venture studio, where we build AI products, tools and workflows. We work hands-on with the tools below. That is how we can tell you which fit your business, which to skip and what each costs to run. You stay free of any one vendor.
- Anthropic
- OpenAI
- Mistral
- Gemini
- Llama
- Hugging Face
- Ollama
- NVIDIA
- Perplexity
- GitHub
- Composio
- Kling AI
- OpenRouter
- DeepSeek
- Qwen
- Kimi
- GLM
- Google Cloud
- Snowflake
- Databricks
- Stripe
- Zapier
- Make
- n8n
- Lindy
- ElevenLabs
- HeyGen
- Midjourney
- Runway
- Higgsfield
- fal
- Apify
- Supabase
- PostgreSQL
- DuckDB
- Spark
- Airflow
- Python
- PyTorch
- LangChain
- Hermes
- OpenClaw
- FastAPI
- Redis
- Kubernetes
- Docker
- Terraform
The people who set your strategy stay to make it work.
Operance AI is an AI consultancy and venture studio with offices in Den Bosch, the Netherlands, and Melbourne, Australia. The team that sets your strategy also puts it to work and stays with it, so accountability never changes hands. Your proposal names which of us answers to you for the result, and that person is still there at the end. Between the two offices we cover European and Australian working hours.
Where we work fromLocal time now
- Den BoschThe Netherlands
- MelbourneAustralia
Peter Sprenger
Arno de Kloet
Jonathan Vandionant
Sion de Jong
Wesley Romeijnders
Alex Brown
Daniel Lee
The questions that come up first.
Answered here, so you can forward this page to whoever else has to sign.
How does working with you usually begin?
With a conversation. We reply to your message within two working days to arrange it. We ask how you work and which decision you want made better. Then a written proposal covers what we will do, what it will deliver, what it costs and how the result will be measured. The first piece of work has a fixed price for a fixed scope. Unless your data has to be produced first, it examines one decision you choose. It works out what improving that decision is worth and what it would take, in numbers your finance team can check. What it produces is yours, whether you continue with us or not.
Who is this for, and who is it not for?
Leadership teams accountable for results, such as margin, cost to serve or cycle time. You may have run a pilot that stalled, be scaling AI now or still be deciding where to start. Working with us starts with one decision to improve and one owner. If you cannot name the decision yet, the conversation can help you find it. If no one can own it yet, we will be here when someone can.
Our data is a mess. Is it too early?
Sometimes, and we will say so. If the data already exists as a by-product of daily work, untidy is workable. If someone would have to be hired to produce it, that becomes the first piece of work.
Do we need AI experts of our own?
No. We bring the AI expertise and work alongside your IT team where you have one. From you we need a leader who owns the result, and time with the people whose work will change.
What does it cost?
The first piece of work has a fixed price for a fixed scope, quoted once we know which decision you want examined. It is the smallest commitment we offer. Every later phase is priced on its own in the proposal, next to the result it is meant to deliver. After each phase you decide whether we continue, and you keep what it produced.
How long before the results show?
If the result you want is already measured today, you can track it from the day AI goes live in your business. If not, we take a starting measurement before anything changes. Either way, your proposal states when the result is first measured after launch. Before you sign, you know when you will see whether it improved.
What if the result does not improve?
The measurement your proposal sets out will show it, and we will say so. You can stop at the end of any phase, and everything we set up is yours, in writing.
Who owns the work and the data?
You do, from day one, in the contract. That covers your data, what we set up for you, its settings, the material any model is trained on and the documentation. Where each part runs is agreed in the contract. When a better AI model arrives, what the system has learned about your business stays with you.
Who runs the AI once it is live?
You choose in writing whether we run it or your team does. If we run it, we answer questions, watch quality, explain costs and replace models when a supplier retires one. Your proposal prices that work. If your team runs it, the settings, the material any model is trained on and the documentation are already yours.
Where does our data run, and how is the AI tested?
Where each part runs is agreed in the contract, and your data is yours from day one. AI that acts on its own can be tricked or misused. Our Security service tests it the way an attacker would, from hidden instructions to data leaving by the wrong route. The gaps found are closed before launch, and every test is on record.
What may the AI decide, and who checks it?
Only what is written down for it. Each AI agent has one named job, written limits on what it may decide and a person who handles the exceptions. Every action is logged, so an audit is a quick search. Our Responsible AI service adds a written record of what the AI decides alone, where a person signs off and how it was tested. The record is ready before anyone asks.
The conversation.We listen first. We ask how you work, where you want to grow and which results you are accountable for. Then we agree which decision to examine first.
The proposal.A written proposal covering what we will do, what it will deliver, what it costs and how the result will be measured. Everything we set up is yours. If you want us to run it and keep improving it once it is live, the proposal prices that too.
The first piece of work.It has a fixed price for a fixed scope. Unless your data has to be produced first, it examines one decision you choose. It works out what improving that decision is worth and what it would take, in numbers your finance team can check. What it produces is yours, whether you continue with us or not.
Tell us a little about your business and what you want to improve. We will reply within two working days to arrange a conversation.