Introduction
This documentation will help you understand and effectively utilize the Siesta AI platform.
What is Siesta AI?
The product specification provides a comprehensive description of the Siesta AI platform, a tool designed for companies that want to fully leverage the potential of artificial intelligence in their daily operations. The platform builds on the capabilities of current large language models but offers a range of key features that are essential for professional and secure deployment in a corporate environment.
Siesta AI enables companies to securely and scalably connect their own knowledge databases with powerful AI models. In addition to connecting to internal documents or databases, it also supports integration with external tools (such as calendars, email, CRM, etc.). The platform can also be controlled directly through the Copilot, which is a chat interface designed for quick control of the platform and agents. Users thus interact with an environment that generates responses based on connected knowledge. Thanks to model independence, any LLM can be utilized, whether in the cloud or on private servers, giving companies full control over where and how their data flows. Built-in analytics and feedback loops allow for monitoring operational metrics, evaluating response quality, and continuously improving agent behavior.
Main Areas of the Platform
The main areas that the platform focuses on are:
Agents
Agents are the fundamental building blocks of the Siesta AI platform. Each agent represents an independent digital agent built on a language model, connected to specific data, equipped with its own instructions, and configured to serve its specific purpose. The platform allows for the creation of countless such agents and their centralized management. Each agent can be assigned to a different team, department, language, or business scenario.
Users can create agents from scratch or from prepared templates, set their behavior, attach datasets, define permissions, and modify their outputs. Agents can be connected to external tools (calendar, email, CRM, etc.) and can perform specific actions either based on user instructions or automatically through a scheduler. The configuration of each agent includes its own system instructions, model choice, output format, creativity level, data access, and options for public or internal deployment.
Each agent has its own detail. This section serves as a central place for managing agents and is divided into several functional tabs covering all areas, from basic information to analytics and feedback to change history.
Agents can be accessed via chat, deployed as a public widget, integrated into Slack, Teams, or other communication channels, or worked with directly through the API. The platform also allows for connecting an agent to specific tools that it can actively utilize.
Integrations, Tools, and Automation
One of the key areas of the Siesta AI platform is the ability to connect agents to external tools, systems, and services. Thanks to this feature, agents are not just a conversational interface but active elements of the digital infrastructure that can perform specific tasks, pass data, or respond to events in real time.
Each agent can be assigned one or more tools that it can use during interactions or automatically in the background. For example, connecting to a calendar, file storage, an internal system via API, or an external webhook. Tools are centrally managed within the Tools interface and can be assigned across agents.
Data
With Siesta AI, it is possible to connect artificial intelligence to the specific knowledge of the organization. Agents do not respond based on a general model but draw information from precisely defined datasets. These datasets can contain both uploaded documents and connections to live systems and knowledge bases that the company uses in its regular operations.
The platform allows for the creation, management, and configuration of so-called datasets - logically separated sets of information that are subsequently assigned to specific agents.
Each agent can be assigned which datasets it is allowed to use, thereby precisely controlling what information it has access to. This approach not only increases the relevance of responses but also ensures a high level of security, as no agent has access to data that has not been explicitly assigned to it.
Smart Chat
The Siesta AI platform provides a unified interface for communication with AI agents, available to both internal users and external visitors. The goal of this part of the system is to create an environment where users can naturally communicate with individual agents, receive relevant responses based on company data, and provide feedback that serves to further improve their behavior.
The platform also includes a so-called smart chat, which is a public anonymous interface that can be embedded on websites or internal portals of the organization. This chat is accessible without login and can be used, for example, for customer support, career communication, or as a tool for answering frequently asked questions. The public chat plugin is fully integrated with the platform, maintains all security principles, allows for feedback collection, and is connected to audit logs.
The platform also supports access to agents via API or integration into commonly used tools like Slack or Microsoft Teams. This allows users to communicate with agents directly from the environment they use daily, without the need to log into a new application. Each agent can be deployed across different channels simultaneously, while the logic, data, and behavior remain consistent.
User and Access Management
The Siesta AI platform allows for detailed management of users, their roles, and access permissions. Each user in the system is assigned a role that determines what functions, data, and agents they have access to.
Administrators can create and modify user accounts, set permission granularity, and precisely define who can modify agents, work with datasets, view feedback, or access system settings.
Access can be controlled not only at the platform level but also within individual components, such as at the level of a specific agent or dataset.
Feedback
One of the key features of the Siesta AI platform is the ability to work with user feedback and continuously improve the behavior of individual agents based on it. Each agent's response can be rated by the user - either as positive or negative. The user can attach a comment to their rating (both positive and negative) in which they explain the clarity, correctness, or usefulness of the response in more detail.
Security and Audit Logs
Every important operation in the system - whether it involves modifying an agent, changing permissions, working with data, or generating responses - is recorded and traceable.
The system includes detailed audit logs that capture who did what, when, and over what. These records are available to administrators in a clear format and allow for retrospective control over all changes in the system.
The platform also allows for the management of API keys, access control through roles, and support for single sign-on (SSO). The security policy meets the needs of organizations that emphasize data control, auditability, and operational reliability.
Analytics and Reporting
The Siesta AI platform includes built-in tools for monitoring traffic, user behavior, and the performance of individual agents. The goal of this part of the system is to provide administrators and managers with an overview of how agents are being used, what impact they have, and where there is room for improvement.
Getting Started
To get started, we recommend going through the Login and Controls sections.
The documentation is continuously updated. For the latest information, please contact us at info@siesta.ai.