# Yada — AI gateway — ALTAI

> Yada applies your organisation’s rules to model requests: keep a request local, mask detected sensitive fields, or block it.

## Yada. Your rules decide what reaches a model.

Yada applies your organisation’s rules to model requests: keep a request local, mask detected sensitive fields, or block it. We configure model access around your data policy.

[Talk to us](/contact/?interest=Yada)

1. **Masks:** National IDs, keys, passwords. The model sees a placeholder.
2. **Blocks:** A request that breaks a rule never reaches a model.
3. **Records:** Which rule, which model. Raw prompts are not stored.

Yada’s gate reads two requests. The first asks for a contract summary for Ayşe Demir with her ID number: the name and the ID number are masked, and the request is let through. The second asks to send the customer list to an external API: it is blocked. Both decisions are recorded; raw prompts are not stored.

### Apply your rules to model requests.

Yada sits between applications and local or cloud models. Configured rules select the model and determine whether detected sensitive data is masked, the request stays local, or the call is blocked.

#### One endpoint, defined model choices

OpenAI-compatible applications connect to a shared model endpoint. Operator rules decide which local or cloud model receives each request.

#### Data handling rules

Configured checks apply allow, redact, local-only or block actions to detected personal data and secrets.

#### Inspectable policy decisions

Operators can see the rule, model and scan action applied to a request. Raw prompts are not logged by default.

### Where it fits.

| Team | The task | What changes |
| --- | --- | --- |
| Enterprise IT | Keep selected requests on local models and permit cloud access under defined conditions. | Manage application model choices and data transfer rules in one place. |
| Internal software teams | Apply a shared policy to model calls from several applications. | Manage changes to models and policies separately from application code. |
| Data and security teams | Define detection and handling rules for sensitive fields specific to the organization. | Test data transfer decisions on real examples and inspect the rule applied. |

### Set up around your work.

ALTAI configures Yada around your data classes, model choices and existing applications. In a solution built with isanagent, Yada can govern model requests and data policy while isanagent runs workflows that use tools. The connection and rules are designed and tested as part of the project.

**We measure together**

- Model selection and policy correctness on agreed scenarios
- Sensitive-field detection and handling on test data
- Request latency and successful completion rate by model

### Spec sheet

- **Category:** Model access and data policy
- **Role:** A foundational ALTAI technology for managing model access and data policy in solutions built for each client.
- **Routing:** Apps call one endpoint, a rule picks the model
- **Interface:** OpenAI-compatible API, one model name
- **Local models:** Ollama, OpenAI-compatible servers
- **Cloud:** Gemini, OpenAI, Anthropic
- **If scanning fails:** The request stays on the local model.
- **Built for:** IT, Software, Information security
- **Works with:** [isanagent](/products/isanagent/)

### Discuss your model and data policy.

[Talk to us](/contact/?interest=Yada)

Source: https://altai.dev/products/yada/
