AI Decisioning Platforms

An AI decisioning platform is the software layer where a business rule meets a machine learning model and something gets decided. Approve the loan or decline it. Flag the transaction or let it through. Route the claim to automatic settlement or to a human. These are high volume, repeatable decisions that a company makes millions of times a day, and the platform is what makes them consistent, fast, and explainable after the fact.

The revealing thing about this category is who wins it. The vendors at the top are not the names most associated with artificial intelligence. They are credit scoring firms and rules engine companies, several of them decades old. That is not an accident of analyst coverage. It reflects what the category is actually selling, which is not intelligence but accountability for decisions that a regulator may eventually ask about.

What the platform actually does

Strip away the branding and these products do four things. They ingest data from wherever it lives. They let someone author decision logic, ideally without writing code. They execute that logic in real time against live data. And they keep a record of what was decided, on what basis, using which version of which model.

The fourth one is the product. Plenty of tools can score a customer. Far fewer can tell you, eighteen months later, exactly why a particular applicant was declined in March, which model version produced the score, who changed the threshold, and whether that change was tested before it went live.

That is why the buying centre for this category tends to sit in financial services, insurance, telecommunications, and government rather than in marketing or product teams. The decisions being automated carry regulatory consequences, and the audit trail is the feature that justifies the licence.

Why a rules engine market got renamed

This category has been through several names. Business rules management. Digital decisioning. Decision intelligence. The underlying software is recognisably continuous across all of them, and some of these vendors have been selling into the same accounts since long before anyone described their product as AI.

What changed is the ratio of hand-written logic to learned logic. A rules engine encodes what a human expert already knows. A decisioning platform blends that with models that infer patterns nobody wrote down. Once both sit in the same execution path, the governance problem becomes considerably harder, because a model drifts and a rule does not.

Agentic AI has pushed this further. An autonomous agent taking action inside a business needs somewhere to make its decisions that is bounded, logged, and reversible. Forrester's framing treats the decisioning platform as exactly that substrate, with modular architecture letting organisations build decisioning agents that adapt to real time data within defined limits. That is a meaningfully different pitch from the one this category made in 2023, and it is the reason a mature market suddenly has new relevance.

Who is in the market

The evaluated set runs to fifteen providers: ACTICO, CRIF, Decisions, Experian, FICO, FlexRule, IBM, InRule, Palantir, Pegasystems, Progress, Provenir, Sapiens, SAS, and Sparkling Logic.

Read that list as a description of the market rather than a shortlist. It splits into roughly three groups. Credit and risk specialists that grew out of lending, including FICO, Experian, CRIF, and Provenir. Horizontal rules and workflow platforms such as Pegasystems, IBM, Decisions, InRule, and Progress. And a set of independents built around decision modelling standards, where Sapiens, Sparkling Logic, ACTICO, and FlexRule sit.

Which group you shop from depends far more on your use case than on tier placement. A lender replacing an origination stack and a telco building next best action logic are buying different products that happen to share a category label.

What The Forrester Wave: AI Decisioning Platforms, Q2 2025 found

The current evaluation is The Forrester Wave: AI Decisioning Platforms, Q2 2025, published on 10 June 2025 by Mike Gualtieri with three contributors, scoring fifteen providers against eighteen criteria across current offering, strategy, and customer feedback. It replaced a Q2 2023 edition that covered thirteen vendors, so the market has grown modestly rather than consolidated.

FICO took the highest score in current offering and the second highest in strategy, with top marks on thirteen criteria, clustered in decision authoring, testing, optimisation, and the governance capabilities around lifecycle, transparency, and observability. Pegasystems took perfect scores across every strategy criterion, including vision, roadmap, and innovation, along with top scores in data modelling and integration. IBM was recognised as a Leader on the strength of decision authoring and optimisation, with the machine learning tooling doing the heavy lifting. Sapiens also placed as a Leader, positioned specifically around regulated industries and no code decision modelling.

Below the Leaders, Provenir entered its first Wave as a Strong Performer, with the report pointing at user experience and an all-in-one credit lifecycle scope covering risk, fraud, identity, collections, and customer management. FlexRule placed as the strongest Contender.

The pattern worth noticing is that scores clustered around governance and authoring criteria rather than model sophistication. Nobody won this evaluation by having better algorithms.

What this means for a shortlist

The questions that separate these products are narrower than the category name suggests.

Ask who can author and change decision logic without engineering involvement, and then ask what happens between someone changing a threshold and that change reaching production. The gap between those two answers is where most decisioning projects fail after go-live.

Ask what the platform records by default rather than what it can be configured to record. Audit capability that requires bespoke work is audit capability you will not have when you need it.

And ask whether the vendor's prebuilt content matches your domain. Several of these platforms carry substantial credit risk libraries, which is enormous leverage if you are lending and irrelevant if you are not. The horizontal platforms make the opposite trade. That single question narrows fifteen names to three or four faster than any tier placement will.

Forrester does not endorse any vendor named here, and tier placement should not be read as a recommendation to buy.