Applications · Enterprise Reasoning in practice
Enterprise Reasoning is a different way to approach AI: start from the decisions your organisation has to make, not the technology it could deploy. You can lead that change by spotting where it matters, shaping the plan and building the capability, one decision at a time.
This page turns the ideas from the book and Learn & Experience into practical steps, using the same supplier decision as an example.
How you build it
These are the three phases of the capability roadmap, applied to one decision class. Each phase has its own job, and alongside each are initiatives you could lead.
Pick a decision class, such as approving suppliers. Look at what makes it hard and how much reasoning it demands, work out the capability it needs (what the book calls required Enterprise Reasoning Capability, or ERC*), and compare that with what the organisation has today. This phase is diagnosis, not implementation.
Build what the decision needs. For supplier approval, that means deciding which people and machines weigh in, how their views are reconciled, where the shared record of the decision lives, who holds authority and which safeguards apply.
Put the new decision system to work. Track decision quality, losses avoided, time saved and lessons learned against what you set out to achieve. Strengthen what works, then carry the approach to other decision classes where it fits.
What you can help your organisation build
Keep these four in view through all three phases. They are ongoing responsibilities, not extra stages.
Make decisions the organising principle and Enterprise Reasoning the lasting capability, whatever happens to models, vendors and technology fashions.
Design how specialist reasoning, a shared record of the decision and the reconciling of views come together, so many inputs become one coherent decision.
Build these into how decisions turn into action: Decision Integrity, clear limits on authority, checks before anything is executed, real human oversight, and the ability to undo what has been done.
Measure how each decision class performs today, track improvements and losses avoided, and use the results to decide what to improve and where to scale.
Where to start
You don't need an organisation-wide technology overhaul. Pick one kind of decision where the stakes are high, it comes up often and there's clear room to do better.
Choose a clear decision class. Name the decision, who owns it, what has to be decided and whose authority allows action.
Make the demand visible. Before choosing any solution, look at its complexity, conflict, uncertainty, velocity, autonomy, scale and consequence. For the supplier switch, conflict and consequence stand out.
Make the case for change. Compare the capability the decision needs with what you have, record how decisions perform today, and pilot where the value justifies the risk.
The goal isn't another collection of separate AI use cases. It's an organisation that gets steadily better at reasoning, one important decision class at a time.
Your first step
Bring a decision from your own organisation into the lab and get its Reasoning Demand (RD) profile.
Want the skills to lead this in your own organisation? Our training programmes cover the framework and how to apply it.
Explore Training Programmes