Decisions API

API for decision models to classify, score, and route text.

1 views this week0 upvotes

About Decisions API

Decisions API is a robust AI customer support platform designed to streamline decision-making processes by integrating multiple AI decision models into a single, unified API interface. It empowers developers to classify, score, and route textual input efficiently, providing structured outputs complete with choices, probabilities, and scores. This API supports simultaneous evaluation of multiple questions, delivering probabilistic answers that can significantly enhance automation in customer support workflows.

The platform comes equipped with a rich playground environment where users can experiment with a variety of decision models such as Jev, Kev, and Span, testing real-life examples with ease. This flexibility allows teams to select the best-suited model for their specific use cases without modifying their backend integrations, thanks to standardized request and response formats. Furthermore, Decisions API offers a credit-based pricing system and provides new users with free starter credits, enabling an accessible entry point for both small-scale tests and large-scale deployments.

Use Cases

Customer support triage: Automatically classify incoming support tickets to route them efficiently to the appropriate service teams, reducing response times and improving customer satisfaction.
Agent routing: Analyze messages to determine and assign the best-qualified agent or department to handle a request, optimizing resource allocation.
Citation verification: Use decision models to validate and verify citations or claims in documents, supporting compliance and fact-checking workflows.
Content moderation: Employ AI decisions to classify and moderate user-generated content rapidly, ensuring platform safety and policy compliance.
Lead scoring: Evaluate and score incoming leads based on textual interaction or submissions to prioritize follow-up and increase conversion rates.
LLM guardrails: Integrate probabilistic decision outputs to enforce safety and quality controls on large language models, preventing unwanted outputs.

Key Features

Access multiple decision models from one API
Structured answers with probabilities and scores
Support for multiple questions in a single request
Playground to test models with examples
Credit-based pricing with free starter credits

Pricing

Decisions API operates on a credit-based freemium pricing model. New users get 100 free credits upon account creation to explore and test the API. Typical pricing for most decision models like Jev and Kev is 600 credits per 1 million input tokens, with upstream input costs around $0.042 and site input costs approximately $0.06 per million tokens. Certain lightweight models, such as the Span-01 Lite behavior checks, charge 1 credit per successful API request with no additional token fees. Users pay based on the model selected and usage volume, allowing scalable and cost-effective integration.

Pros & Cons

Pros

+ Unified access to numerous decision models through a single API endpoint simplifies integration and reduces development overhead.
+ Structured responses include choices, probability estimates, and scores, enabling nuanced decision-making and better automation.
+ Support for multiple questions in one request improves efficiency by reducing the number of API calls.
+ Interactive playground facilitates testing and experimentation with models before integration, optimizing accuracy and fit for needs.
+ Credit-based pricing with free starter credits lowers the entry barrier for new users and supports scalable usage.

Cons

- Pricing depends on model and input volume, making costs potentially variable and requiring monitoring for budget control.
- While well-suited for text input, the API’s full potential with image input models is less emphasized, which may limit multimodal use cases.
- Certain models may have usage restrictions or require specific understanding of their scoring outputs, adding some learning curve.
- API integration requires server-side setup and authentication, potentially limiting availability for no-code users.

Frequently Asked Questions

What is Decisions API primarily used for?

Decisions API is mainly used to automate decision-making workflows in customer support and related fields by processing textual inputs to classify, score, and route requests through AI decision models.

Can I try Decisions API before purchasing credits?

Yes, new users receive 100 free credits to explore and test the API models in the playground and through API calls before committing to paid credits.

How does Decisions API handle multiple questions in one request?

The API supports querying multiple decision questions simultaneously within a single request, returning structured answers for each question alongside probabilities and scores.

Are there different models available and can I switch between them easily?

Yes, the platform offers multiple decision models such as Jev, Kev, and Span. You can switch between models seamlessly without changing your backend integration due to standardized request and response formats.

What pricing model does Decisions API use?

Decisions API uses a credit-based pricing system where different models and usage volumes consume credits at varying rates, with typical pricing around 600 credits per million input tokens.

Is Decisions API suitable for non-text inputs like images?

While primarily designed for text inputs, Decisions API also supports some models that handle embedded images, although its focus remains on text-based decision workflows.