What is decision intelligence?
Last reviewed 7 October 2026
Decision intelligence is a discipline that treats business decisions as things to be designed, recorded, measured and improved, using data, analytics and AI to support or automate them. The term has no single settled meaning. Gartner, as quoted by the vendor Aera Technology, calls it a practical discipline for understanding and engineering how decisions are made and how outcomes are evaluated and improved by feedback.
Also called: DI, decision intelligence platform, decision engineering, decision automation.
Why Decision intelligence matters
Most analytics stops at a chart. Someone looks at it, makes a call in a meeting, and the link between the data, the choice and the result is lost. A month later nobody can say whether the decision was good or the outcome was luck. Decision intelligence asks organisations to keep that link: what information fed the decision, what options were considered, what was chosen and what happened.
The term matters to buyers because it is now a product category. Aera Technology's site reports a 2026 Gartner Magic Quadrant for Decision Intelligence Platforms, and quotes Gartner's description of such platforms as software that supports, augments and automates decisions by humans or machines, built from data, analytics, knowledge and AI. Vendors in the category differ widely, so the label alone tells you little.
How it works
The starting point is to name a decision explicitly. Instead of a dashboard for inventory, you define the decision: how much of each product to reorder each week. You record its inputs (stock levels, forecast demand, supplier lead times), its options, its constraints and who owns it.
Next comes the logic. Some decisions can be rules; some need a forecast or an optimisation model; some need a recommendation that a person accepts or overrides. Decision intelligence products typically let you model this flow, run it at scale and decide how much is automated and how much goes to a person.
The loop closes with feedback. Each decision is logged with its inputs and the option chosen, and the outcome is measured against it later. Over time the organisation can see which rules and models produce good outcomes and which do not, and change them. Without this measurement step, a system that calls itself decision intelligence is a dashboard with a new name.
Worked example: weekly reorder decisions
A retailer treats reordering as a decision to design. For each product, the system reads stock, a demand forecast and the supplier's lead time, and proposes an order quantity. Proposals below a value threshold go through automatically. Proposals above it go to the category buyer, who sees the inputs and the reason, and accepts, edits or rejects them.
Every proposal, override and reason is stored. Eight weeks later the team compares outcomes: products where buyers overrode the proposal had fewer stock-outs in one category and more waste in another. The team adjusts the forecast for the first category and tightens the threshold for the second. The improvement comes from the record of decisions, not from a better chart.
How Swfte relates to it
Designed for
Swfte is not a decision intelligence platform and does not offer forecasting or optimisation models. The Swfte Intelligence Platform is designed around a related loop: connect data, analyse, visualise, decide, build, govern, measure, learn. The analysis views and the step that turns a finding into an agent or a workflow are design intent, not shipped features.
What is built today sits on either side of the decision. Studio workflows can route a proposed action to a named person through an approval step, and the approve or reject result is recorded with the run. The company brain records facts about the organisation with an evidence status, so a decision can show how sure the data behind it is.
Related terms
- Business intelligence tools
Business intelligence tools are software for collecting, modelling, analysing and presenting an organisation's data as reports, dashboards and visual analyses that inform decisions.
- Semantic layer
A semantic layer is software that sits between stored data and the tools that query it, translating tables and columns into named business terms such as revenue, active customer or churn.
- AI automation
AI automation is the use of AI models, such as language models or classifiers, inside automated processes to handle steps that fixed rules cannot, like reading a free-text email, classifying a document or choosing the next action.
- Human-in-the-loop (HITL)
Human-in-the-loop (HITL) refers to a system in which a person takes an active part in the operation, supervision or decisions of an automated process at defined points.
- Knowledge graph
A knowledge graph is a store of facts in which entities, such as people, products or systems, are nodes and the named relationships between them are edges.
Common questions
- Is decision intelligence the same as business intelligence?
- No. Business intelligence reports and visualises what happened. Decision intelligence focuses on a specific decision: its inputs, the logic that produces a choice, who acts on it, and measuring the outcome so the logic can improve. Many decision intelligence products include BI-style views, but the decision record is what sets them apart.
- Who coined the term decision intelligence?
- The term has been used by several people and firms, and accounts differ. Gartner defines it and evaluates decision intelligence platforms in a Magic Quadrant. Rather than rely on one origin story, check how each vendor or analyst defines it, because the definitions do not fully agree.
- Does decision intelligence mean decisions are automated?
- Not necessarily. Gartner's description, as quoted by Aera Technology, covers supporting, augmenting and automating decisions. Many designs automate routine, low-value choices and send larger ones to a person with a recommendation and its reasons. How much to automate is itself a decision the owner should record.
- When is decision intelligence overkill?
- When a decision is rare, low-stakes or already made well by an experienced person, modelling it formally costs more than it returns. It pays off for frequent, repeatable decisions with measurable outcomes, such as pricing, reordering or credit limits, where small improvements add up.
Sources
Definitions on this page were read on the sources below on 7 October 2026. Where sources define the term differently, the page says so. The full glossary lists more terms.