A software system that perceives its environment, makes decisions and takes actions to achieve a goal — without requiring a human to approve every step. Gen 3 AI agents can reason, use tools and remember context across interactions.
Agentic AI
AI systems that operate with a degree of autonomy — planning, deciding and acting over time to achieve goals. Synonymous with Gen 3 agents.
AI guardrails
Rules and filters that constrain what an AI agent can say or do — preventing harmful outputs, off-topic responses or policy violations.
C
Context window
The amount of text (measured in tokens) an LLM can process in a single interaction. Larger context windows allow longer conversations and documents.
Conversational AI
AI designed for dialogue. Covers chatbots (Gen 2) and voice agents. ShiftMate's Gen 3 agents are conversational AI plus tool use and memory.
E
Embedding
A numerical representation of text (or other data) that captures its meaning, allowing AI systems to search by semantic similarity rather than exact keyword match.
F
Fine-tuning
Continuing to train a pre-trained model on your own data so it specialises in your domain, style or task. Expensive but powerful for specific use cases.
G
Gen 3 AI
The third generation of AI systems. Gen 1 = rule-based scripts. Gen 2 = LLM chatbots. Gen 3 = agents that reason, remember context, use tools and act autonomously on multi-step tasks.
H
Hallucination
When an AI confidently states something factually incorrect. A known failure mode of LLMs that RAG and fine-tuning help reduce.
I
Inference
Running an AI model to generate a prediction or response. Inference costs (per call) are what you pay when using AI APIs.
L
LLM (Large Language Model)
A neural network trained on massive amounts of text. LLMs power most modern AI assistants and are the reasoning engine inside AI agents. Examples: GPT-4, Claude, Gemini.
M
Multi-modal
AI systems that can process multiple types of input — text, images, audio, video — rather than text alone.
O
Orchestration
The layer that coordinates multiple AI models, tools and data sources to complete a complex task. ShiftMate Gen 3 uses a proprietary orchestration layer.
P
Prompt engineering
The practice of crafting and optimising the instructions given to an LLM to get consistent, accurate outputs. A key skill for getting reliable behaviour from Gen 2 systems.
R
RAG (Retrieval-Augmented Generation)
A technique where an AI retrieves relevant documents or data before generating a response. Prevents hallucination and keeps answers grounded in your specific business knowledge.
S
SLA (Service Level Agreement)
In AI deployments: the contractual commitment to uptime, response time and performance. Important when evaluating AI vendors for production use.
T
Tool use / function calling
The ability of an AI agent to call external systems — APIs, databases, CRMs, calendars — and act on the results. This is what separates Gen 3 agents from Gen 2 chatbots.
Token
The basic unit of text processed by an LLM — roughly ¾ of a word. API pricing is typically charged per token.
V
Voice AI
AI systems that interact through spoken language — listening, understanding and responding in real time. Covers speech-to-text, language understanding and text-to-speech.
Vector database
A database optimised for storing and querying embeddings. Used in RAG pipelines to find semantically similar content quickly.