Skip to content
API coming soon

Decisions API.
A clear next step.

Decisions API is an upcoming decision-making API for software that needs a choice, a score, or a yes/no answer. Turn application context into structured decisions your workflow can use.

Explore the concept

In development · Public access is not yet available

decision flowConcept illustration

01 / Context

“My invoice looks wrong. Can someone help?”

02 / Question

Which team should handle this?

03 / Decision

billing_support

Illustrative outcome only. This is not a live response or a published API contract.

Built around the decision

What is a decisions API?

A decisions API is an interface that lets an application submit context and receive a defined judgment. Instead of leaving a developer to interpret a paragraph, the result can fit a specific branch in the software: choose a category, assess a condition, or assign a score.

The term covers several approaches. Traditional decision services apply business rules or decision tables; newer AI decision interfaces evaluate questions about supplied content. The useful common ground is a clear boundary between the information you send, the decision you request, and the action your application takes.

Yes or no

With a decisions API, evaluate a specific condition, such as whether a message needs a human response. A binary decision gives your application a clear branch to follow, while uncertain cases can remain in a review queue.

Choose an option

A decisions API can select from a set of labels you define: billing, technical support, account access, or another destination. Constrained choices make the result easier to validate and connect to an existing workflow.

Score and prioritize

Use a decisions API to assess an item against a stated scale, such as urgency or relevance. Scores become useful when your team defines what each range means and checks those boundaries against representative examples.

Where decisions belong

Small judgments. Useful workflows.

These are common decision API use cases, offered as planning examples rather than available features of our upcoming service.

01

Support ticket routing

A decisions API can read the customer message alongside useful account context, then assign a category and urgency level. Your support system can route clear cases automatically and reserve ambiguous requests for a person. Keep the original message available so reviewers can understand the situation.

02

Content review

A decisions API can evaluate submissions against explicit editorial criteria before they enter a publishing queue. Decisions might flag missing information, identify an unsuitable topic, or request additional review. The application should treat the result as a review signal, with a defined path for correcting mistakes.

03

Agent workflow gates

Use a decisions API to check whether an AI agent has enough context to proceed or should ask for clarification. A decision can help select the next step, but your application must still enforce permissions. Model output alone should never grant access or authorize an irreversible action.

04

Operational triage

Use a decisions API to sort incoming requests, documents, or alerts into a manageable queue. A decision service can suggest priority based on criteria your team supplies. Validate those criteria against real examples, and keep human review available when an incorrect result would carry a meaningful cost.

From context to action.

STEP 01

Describe the situation

For a decisions API, supply the relevant application state: a message, an object, or a concise set of facts. Include the information needed for the judgment, but avoid unrelated records. Clear context helps distinguish a routine request from one that needs special handling.

STEP 02

Define the decision

For your decisions API request, specify the question, allowed options, or scoring criteria. A narrow question such as “Which support team should handle this?” is easier to evaluate than a broad request to solve everything. Decide how your application will represent an unknown or inconclusive answer.

STEP 03

Validate and act

Check the returned structure before using it. Apply your own business rules, thresholds, and authorization checks, then route the outcome to the appropriate workflow. Record enough context to investigate unexpected behavior and compare results when the decision configuration changes.

Plan for uncertainty.

Before integrating any decisions API, define success with a small evaluation set. Include ordinary inputs, ambiguous cases, and examples that should trigger review. Compare outcomes with your existing process rather than assuming a structured response is automatically correct.

Ask providers how they handle invalid input, timeouts, version changes, data retention, and authentication. Keep decision logic separate from side effects so you can test a judgment without sending a message or changing a record. These are integration considerations, not claims about features already implemented here.

Decisions API questions, answered.

Is Decisions API available now?

Not yet. This website is an introduction to our upcoming decisions API. Public endpoints, API keys, and developer documentation are not available. A release date has not been announced; confirmed access details will be published here when the API opens.

How is a decisions API different from a chat API?

A chat API typically produces conversational text. A decisions API focuses on a constrained judgment, such as a label, score, or yes/no answer. Some chat systems can also return structured output, so the practical difference lies in the interface, validation, and behavior your application needs.

Does a decision API always use AI?

No. Decision services can evaluate deterministic business rules, decision tables, predictive models, or AI models. The phrase describes an interface for making or managing decisions, rather than one specific technology. Our intended focus is structured decisions for software workflows; implementation details will be confirmed at launch.

Can a confidence score guarantee a correct decision?

No. A probability or confidence value needs interpretation in the context of the model and task. Test performance on examples from your own workflow, measure incorrect outcomes, and set review thresholds accordingly. A high score should not replace independent checks for sensitive or consequential actions.

Will pricing and integration examples be published?

Yes, the launch information is intended to explain access, supported decision types, request and response formats, and pricing. Those details are still being defined. This page does not promise a particular endpoint, SDK, model, service level, or free tier before the public release.

Further reading

Background sources for the concepts on this page. These organizations are independent of Decisions API; their documentation does not describe our service.

B.AI: Decisions API reference

An example of an API designed for classification, scoring, and yes/no judgments.

EmpirioLabs: decision models

An example of typed questions and probability-based decision outputs.

OMG: Decision Model and Notation

The standards-based approach to representing business decisions and rules.

Coming soon

The next step is on its way.

We are preparing Decisions API for public access. Visit this page again for launch details, documentation, and confirmed capabilities.