Glossary

Quick reference for terms you will encounter across the Kindo documentation.

Term	Definition
Agent	A Kindo-managed automation that executes a series of steps. Types include Chatbots, Workflow Agents, and Trigger Agents.
Agent Run	The final output of an agent’s last LLM Step.
Agent Step	An atomic operation within an agent: LLM Step, Action Step, or API Action Step.
Auth Backend	The component managing enterprise SSO and authentication. Self-Managed Kindo uses SSOReady.
Chat	A collection of requests from a user to LLMs through the Kindo Terminal without an agent.
Chatbot	An agent that responds to users using information from an AI model and a knowledge base.
DeepHat	Kindo’s proprietary cybersecurity-focused LLM, optimized for DevSecOps tasks and adversary simulation.
DLP (Data Loss Prevention)	Filters that tokenize or redact sensitive data before it reaches AI models.
Embeddings Model	A model that converts content into numeric vectors for semantic search and retrieval.
Environment	(Self-Managed) The Kubernetes platform where Kindo’s Helm Chart runs.
File	A file within the Kindo Platform — uploaded by users, provided by integrations, or generated by AI.
Integration	An external system connected to Kindo for data ingestion and outbound actions (e.g., Jira, Slack).
Kindo API	REST API endpoints: Inference API (single LLM request) and Agent API (agent run output).
Kindo SaaS	The cloud-hosted version of Kindo, managed by the Kindo team.
Kindo Terminal	The primary user interface for making requests to AI models.
LLM	Large Language Model — the AI model used in requests (e.g., GPT-4, Claude, DeepSeek R1).
LLM Step	An agent step that requests a response from an LLM.
Audit Log	A record of user, AI agent, and administrative actions within Kindo. Stored in the database by default; optionally forwarded to syslog in Self-Managed deployments. See Security Controls.
Log Backend	Where Kindo streams its audit log (Syslog server in Self-Managed).
MCP	Model Context Protocol — an open standard for connecting AI agents to external tools.
Memory	The historical collection of LLM responses within a chat or agent session.
Peripheries	Third-party services Kindo depends on (Unleash, Qdrant, Presidio, etc.).
Redis Streams	The Redis data structure Kindo currently uses for real-time conversation streaming between backend services. Requires a non-sharded Redis deployment. See Infrastructure Requirements.
Resource	A structured data entity from an integration (Tickets, Users, Emails, Logs).
Run	An instance of interaction between a user and an LLM brokered by Kindo.
Sandbox	An isolated VM created per agent conversation for secure code execution.
Secret	A sensitive value (API key, token, etc.) stored in the Secrets Vault.
Self-Managed Kindo (SMK)	Kindo deployed in customer-controlled infrastructure via Kubernetes Helm Chart.
Self-Managed Model	An AI model hosted and managed by the customer rather than a cloud provider.
State DB	The backend relational database (PostgreSQL) holding Kindo’s application state.
Step Run	The individual response of a single LLM Step within an agent.
Sync Backend	The component that syncs Kindo with external SaaS services (ticketing, storage, etc.).
Target	The output destination of an agent run (Terminal, ticketing system, API endpoint).
Trigger Agent	An agent that activates in response to a specific event in an integrated system.
Unleash	Feature flag management platform used by Kindo for model configuration and feature toggles.
User LLM	An LLM that users or agents make requests to. Shown as “Models” in the Kindo UI.
Vector DB	Vector database used for RAG and semantic search (Pinecone or Qdrant).
Worker LLM	An LLM used internally by Kindo for non-user operations.
Workflow Agent	An agent that conducts operations within systems of record, including API calls.