Gerwin AI API

1. What is currently available

In short: the public Gerwin AI API currently works only with the previous version of the Service and its skills. New models and applications from the current Gerwin AI interface are not yet available through this API.

This is a supported legacy API for integrations that need existing text-processing skills. Access is limited, reviewed manually, and provided upon application.

A feature, model, or application available in the new web interface is not necessarily available through the API. The current list of skills, endpoints, request formats, limits, and pricing is specified in the technical documentation issued upon access.

2. Skill versus model

A skill is a ready-made operation with a predefined task, input parameters, and output format. For example, a skill may accept source text and perform the transformation defined for that skill.

A model is a particular AI model selected in the new version of Gerwin AI. The legacy API does not provide direct selection of new models or access to the current application catalog.

The basic flow is:

1. the integration selects an available skill;
2. it sends the input supported by that skill;
3. it receives a job identifier;
4. it checks the job status and retrieves the result.

If you need API access to a specific new Gerwin AI model or application, contact support first to confirm availability.

3. Requesting access

Before connection, you must top up the account balance by at least 1,000,000 credits.

A balance top-up does not activate API access automatically and does not guarantee approval. After topping up, submit a request to support. Access is enabled manually after the use case is reviewed.

Email team@gerwin.ai and describe:

• the email address of the funded Gerwin AI account;
• the organization or project;
• intended use case and content types;
• expected request volume and concurrency;
• required skills and output formats;
• data-retention and support requirements.

We may request additional information for safety, capacity, and contract review. Once approved, you will receive connection details and the current technical documentation.

For a first-time integrator, the process has three steps:

1. top up the account balance by at least 1,000,000 credits;
2. email support with a description of the integration;
3. wait for manual review and receive access.

4. Authentication

Use only the authentication method specified in the documentation issued upon access. If a token is obtained through a separate sign-in endpoint, send credentials only to the official Gerwin AI HTTPS authentication endpoint and use the resulting token for operational requests.

Never embed a password or token in a client-side application, browser code, mobile bundle, or public repository.

Recommended controls include:

• keeping secrets on a protected server or in a secret manager;
• using a separate key per environment;
• granting least privilege;
• rotating keys and revoking compromised keys promptly;
• excluding secrets from logs, analytics, and error messages;
• applying source-address and budget restrictions where available.

The customer is responsible for use of an active token or other integration credential until Gerwin AI is notified of compromise.

5. Jobs and outputs

Legacy API operations may be asynchronous. Integrations should handle:

• job acceptance and its identifier;
• intermediate state;
• completion, moderation rejection, and provider failure;
• webhook redelivery;
• temporary unavailability and safe retries.

Do not depend on undocumented fields or object ordering. Use only the skills and parameters listed in the issued documentation.

6. Limits and reliability

The API may apply concurrency, rate, file-size, budget, timeout, and queue limits. Unless a different limit is agreed during onboarding, use the baseline of no more than one request every four seconds. On a rate-limit response, retry with exponential backoff and jitter. Aggressive retries, limit circumvention, and spreading traffic across accounts are prohibited.

Experimental methods may change. For critical integrations, agree versioning, notice, and support terms in a separate contract.

7. Billing

Connection requires an advance balance top-up of at least 1,000,000 credits. The account must maintain a positive balance while the API is in use.

Pricing depends on the selected skill, its parameters, actual volume, and access terms. The legacy API is not billed as direct access to new models because such access is not currently provided.

If a technical failure causes an incorrect charge, send the operation identifier to support. Monetary refunds follow the Refund Policy; internal units may be restored automatically or after review.

8. Moderation and acceptable use

API access does not waive safety rules. Prompts, inputs, and outputs may be checked by Gerwin AI and underlying technology providers.

The integrator must:

• comply with the Content Moderation Rules;
• avoid giving unauthorized third parties direct access to its credentials;
• implement controls appropriate to its product and audience;
• disclose synthetic content where required;
• address end-user abuse promptly.

9. Data and confidentiality

Do not submit personal data, secrets, biometric data, or legally protected information unless necessary and supported by lawful grounds, processing terms, and safeguards.

Corporate processing may require a separate data-processing agreement. General terms appear in the Personal Data Processing Policy.

10. Suspension

Access or a token may be restricted for suspected compromise, exceeded limits, non-payment, rule violations, threats to the Service or third parties, or a legal or critical-provider requirement.

Where possible, we notify the integration owner and allow remediation. An urgent risk may require immediate revocation.

11. Support

Include the organization, integration identifier, timestamp, method, response code, and operation identifier. Never send the full secret, password, card details, or unnecessary user content.

Contact: team@gerwin.ai.