> For the complete documentation index, see [llms.txt](https://docs.siit.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.siit.io/core-plateform/ai.md).

# AI

Siit ships AI as three distinct surfaces, each with its own scope and audience. You decide what every agent can read, where it can act, and which steps need approval

Siit's AI is built to save minutes on every request while keeping humans in control. You decide what data AI can read, where it can act, and which steps require explicit approval. Every suggestion, decision and action is logged on the request timeline.

{% content-ref url="/pages/vMWgzWh1GwYOEu8NTvcN" %}
[AI trust model](/ai/ai-trust-model.md)
{% endcontent-ref %}

#### Three AI surfaces

Siit's AI ships as three distinct products, each with its own scope and audience.

| Agents                                  | Where it runs                                | Who it serves        | What it does                                                                                                             |
| --------------------------------------- | -------------------------------------------- | -------------------- | ------------------------------------------------------------------------------------------------------------------------ |
| [**Assist**](/ai/assist.md)             | The admin dashboard                          | Admins and agents    | Builds your configuration, answers product questions from the documentation, and analyses your live request data         |
| [**Triage Agent**](/ai/triage-agent.md) | Slack, Microsoft Teams, employee Portal      | Employees            | Handles intake conversationally: deflects with a knowledge article when one exists, otherwise creates a complete request |
| [**IT Agent**](/ai/it-agent.md)         | Slack, Microsoft Teams, the request timeline | Employees and agents | Runs the playbooks you publish: real actions in your connected tools, gated by your approval rules                       |

#### Assist: configure and query your workspace by conversation

* **Build my setup**: creates and edits services, service forms, workflows, inboxes and knowledge articles. "Create a service catalog based on past requests" turns an empty workspace into a working one in a single prompt.
* **Ask the docs**: answers product and setup questions from [docs.siit.io](https://docs.siit.io), [help.siit.io](https://help.siit.io) and the API reference.
* **Analyze requests**: reads your live request data and turns it into answers, summaries and drafts, for example "write my inbox weekly update from last week's requests" or "find open requests older than 7 days with no assignee".
* Runs with your role and your permissions: it cannot read anything you could not open yourself, and it pauses for your approval before every write.<br>

#### Triage Agent: intake without a form to fill

* Checks your knowledge base first and suggests the relevant article with a summary. If the article resolves the problem, no request is created.
* Reads your Service Catalog to route each request to the right service.
* Collects every required form field through conversation, including custom and multi-select fields, and validates them against your form's constraints before the request is created.
* Sets priority and applies tags, derived from the conversation content and your service configuration.
* Reads the files an employee shares during intake (screenshots, documents, spreadsheets) and uses them as triage context, so a screenshot of an error can match the service and pre-fill the description.
* Attaches the full AI exchange to the request, so admins get the whole story without opening Slack.

You choose whether the agent creates the request directly or always shows the employee a pre-filled form to review first.<br>

#### IT Agent: action across your tools

* Playbooks written in plain language: you describe when the playbook starts and the steps to follow, and insert actions inline.
* Native actions in Okta, JumpCloud, Google Workspace, Slack, Jira and Zendesk, plus custom webhooks for anything else in your stack.
* Autonomy set per action: **Auto-run** or **Approval required**, with the approver resolved dynamically (the requester's manager) or set explicitly.
* Progressive rollout: publish as **Draft**, then to a pilot audience, then **Live**. Simulate mode logs what would have happened without changing anything.
* Every step (prompt, decision, action, approval) is logged and replayable.

> **Note:** approvals are mandatory for actions that change a third-party system unless you explicitly set them to auto-run. Autonomous resolution is something you grant playbook by playbook, not a default.

#### Pricing plan

|                  | Plan                                                  |
| ---------------- | ----------------------------------------------------- |
| **Assist**       | All plans, with a fair use policy applied per account |
| **Triage Agent** | Pro, with a 14-day free trial available on any plan   |
| **IT Agent**     | Pro, with a 14-day free trial available on any plan   |

### Explore Siit AI agents

{% content-ref url="/pages/52OMx5tN7Qg1PcXolcyF" %}
[Assist](/ai/assist.md)
{% endcontent-ref %}

{% content-ref url="/pages/4oVJ6LnPZPXbfnCNLi9A" %}
[Triage Agent](/ai/triage-agent.md)
{% endcontent-ref %}

{% content-ref url="/pages/OlryuObeb41k838DG7O3" %}
[IT Agent](/ai/it-agent.md)
{% endcontent-ref %}