Core Workflow
DecisionsApi follows a four-step process to turn context into actionable results for your code:
Supply context: Pass any relevant message, policy, record, or application state that the decision needs.
Define the question: Name the decision and its allowed answers, ordered scale, or yes/no criteria.
Inspect the result: Access the typed answer, probability signals, model details, usage data, and full request metadata.
Choose the next step: Use your own application rules to route, score, block, or flag items for human review.
Supported Decision Types
The platform returns three standardized output formats for all models:
Choice: Select one label from a user-defined set, ideal for routing, categorization, filtering, and next-step selection in agent workflows.
Score: Measure urgency, relevance, or completeness on an explicit ordered scale that you define and test.
Noul / yes-no: Ask a focused true-or-false question, and use the returned probability with your own custom thresholds and review paths.
Common Use Cases
You can implement these common production tasks directly with DecisionsApi:
Route customer requests: Distinguish billing, technical, account, and other support issues, then pass the classified label to your existing support system.
Choose an agent's next step: Classify whether a request needs a record lookup, LLM call, tool invocation, or human review before any action runs.
Check claims against evidence: Compare a stated claim to a source passage, and confirm if the supplied evidence supports, contradicts, or cannot verify the claim.
DecisionsApi includes the following capabilities for developers:
Unified API interface that works across all supported models including Jev, Laya, Kev, Solar, and Span
Full structured responses that are REST-ready with no unstructured free text output
Server-side Bearer token authentication with manageable API keys
Online workbench where you can select a task, edit input, define typed questions, and preview requests before connecting to your backend
Unlimited free use of the online playground
Shared credit balance that works across the playground, live API calls, and evaluation workflows
Full request history and credit activity tracking, with request identifiers to connect API calls to their usage records
Pricing Plans
All plans use non-expiring credits that do not expire after purchase:
Starter ($10): 100,000 credits, 1 workspace, 3 concurrent requests, standard speed, access to all three question types, typed output, probability results, online playground, API key management, and email support.
Pro ($100): 1,000,000 credits, unlimited workspaces, 10 concurrent requests, fast lane processing, parallel questions per request, full usage and request history, and priority support.
Enterprise ($1,000): 11,000,000 credits (10% extra included), unlimited concurrency, dedicated fast lane, team workspaces and collaboration tools, custom integration support, security and permission guidance, dedicated support, priority production processing, and product roadmap feedback access.
Current Status
DecisionsApi is in limited invite-only testing. New users receive initial testing credits when they create an account. Access is granted in payment order, and users are notified by email when their account is activated. All one-time plans do not auto-renew and do not require a long-term subscription.
Comments