Name: Utomatic
Address: Rotterdam, NL
Telephone: +31850041088
Price range: €€€

AI Basics

Artificial Intelligence (AI): Computer systems that perform tasks normally requiring human intelligence, such as learning, reasoning, and understanding language.
Machine Learning (ML): A branch of AI where systems automatically improve through experience, without being explicitly programmed.
Deep Learning: A form of machine learning based on neural networks with multiple layers, inspired by the human brain.
Neural Network: A mathematical model inspired by the human brain, consisting of layers of connected nodes (neurons).
Natural Language Processing (NLP): The ability of AI to understand, interpret, and generate human language in both text and speech.
Generative AI: AI that creates new content — text, images, code, or audio — rather than just analyzing existing data.

AI Models & Technology

LLM (Large Language Model): AI models trained on massive amounts of text, such as GPT-4, Claude, and Gemini. Can understand and generate text.
GPT: A family of language models developed by OpenAI, including ChatGPT. GPT stands for Generative Pre-trained Transformer.
Transformer: The neural network architecture that powers modern LLMs. Introduced by Google in 2017.
Parameters: The internal variables an AI model learns during training. More parameters often (not always) means a more capable model.
Token: The basic unit of text an LLM processes — usually part of a word. 1 token ≈ 4 characters (exact ratio depends on the tokenizer model).
Context Window: The amount of text an AI model can process at once. Larger context windows (100K+ tokens) can handle entire documents.
Fine-tuning: Specializing an AI model on your specific data or task, on top of its general training.
Embedding: A numerical representation of text that lets AI compare meaning. The foundation of semantic search.
RAG (Retrieval Augmented Generation): AI technique where the model retrieves relevant info from your business documents before answering. Ensures accurate, source-specific responses.
Vector Database: A database specifically for storing and searching embeddings. Essential for RAG systems.

AI Agent: An autonomous AI that performs tasks independently — thinking, deciding, using tools — rather than just answering questions.
Chatbot: An AI program that communicates with users in natural language via text. Usually for customer service and lead generation.
Voice Agent: An AI that can conduct phone calls with a natural voice, schedule appointments, and qualify leads.
RPA (Robotic Process Automation): Software robots that automate repetitive computer tasks like data entry, email processing, or invoicing.
Workflow Automation: Automating business processes by connecting multiple tools and systems together.
Copilot: An AI assistant that helps humans do their work faster, without fully operating autonomously. Examples: GitHub Copilot or Microsoft 365 Copilot.
Function Calling: An LLM's ability to call external functions or APIs, like checking a calendar or updating a database.
Multi-agent Systeem: Multiple AI agents working together, each with their own specialty, to solve complex tasks.

Prompt: The instruction or question you give to an AI model. Better prompts produce better results.
Prompt Engineering: The craft of writing effective prompts to get optimal AI results.
System Prompt: Background instructions that determine an AI's behavior, such as personality, role, and rules.
Hallucinatie: When an AI generates convincing-sounding but factually incorrect information.
Temperature: A setting that controls how creative or predictable AI output is. Low temperature = consistent, high temperature = creative.
Chain-of-Thought: A technique where AI reasons step-by-step before giving an answer. Improves accuracy.
Few-shot Learning: Giving an AI examples in the prompt so it learns what you want without fine-tuning.
Zero-shot: Having an AI perform a task without examples — based on instructions alone.
Grounding: Connecting an AI to real data (documents, databases) to give factually correct answers.

ROI (Return on Investment): The return on an AI investment. E.g., an AI voice agent costs €5K but saves €20K/year = 4x ROI.
MVP (Minimum Viable Product): A first working version of an AI solution to test before full investment. Similar to our pilot.
Pilot: A small-scale test of an AI solution in your business, usually 1-2 weeks, to prove value before scaling.
API Integratie: Connecting AI to your existing systems (CRM, calendar, email) via programming interfaces (APIs).
Human-in-the-Loop: System design where AI prepares decisions but humans give final approval. Essential for high-risk tasks.
Time-to-Value: The time between implementation and measurable business results. Utomatic delivers this within 2-4 weeks.
SLA (Service Level Agreement): Contractual agreements about performance, e.g., uptime guarantee or response time.

AVG / GDPR: European privacy law governing how companies process personal data. All Utomatic AI solutions are GDPR-compliant.
EU AI Act: European AI legislation (2024) that classifies AI systems by risk and imposes corresponding obligations.
Bias: Systematic skewed judgments by AI, often due to biased training data. Important to mitigate.
Data Sovereignty: The principle that data is subject to the laws of the country where it's stored. Utomatic uses EU servers.
Verwerkersovereenkomst: Contract between a company and AI provider about how personal data is processed. Required under GDPR.
Explainability: The ability to explain why an AI made a particular decision. Required for high-risk applications.

Speech-to-Text (STT): Technology that converts spoken words into written text. Foundation of voice agents.
Text-to-Speech (TTS): Technology that converts text into natural-sounding speech. Modern TTS is barely distinguishable from humans.
Voice Cloning: Creating an AI copy of someone's voice based on a short recording.
Intent Recognition: AI that understands what a user wants to achieve, regardless of exact wording.
Semantic Search: Search method that understands meaning, not just exact words. "Car" also finds "vehicle" or "automobile".