AI Glossary for Contractors: 50+ Terms Explained

A

AI (Artificial Intelligence)

Software that can perform tasks that normally require human intelligence — things like understanding language, recognizing patterns, making decisions, and learning from experience. AI isn’t one technology; it’s an umbrella term for many different approaches. When a tool claims to use “AI,” it could mean anything from simple rule-following to sophisticated pattern recognition. Contractor example: An AI phone answering system that understands a caller saying “my AC broke” and asks the right follow-up questions — not because someone programmed every possible response, but because it learned how those conversations work.

Algorithm

A set of step-by-step instructions that a computer follows to solve a problem. Think of it like a detailed scope of work — it tells the system exactly what to do, in what order, under what conditions. Every AI tool runs on algorithms, though the sophisticated ones are so complex that even their creators can’t always explain every step. Contractor example: A scheduling algorithm that assigns techs to jobs based on location, skills, availability, and priority — running through every possible combination to find the best assignments.

API (Application Programming Interface)

A way for software programs to talk to each other. When your CRM sends job data to your accounting software automatically, that connection happens through an API. You don’t need to understand how APIs work technically — you just need to know that when a tool says it “integrates via API,” it means it can share data with other tools without you doing manual entry. Contractor example: Your AI phone system captures a lead and automatically creates a record in your CRM through an API — no human copies and pastes anything.

Automation

Making a process run without human intervention. Automation follows fixed rules — “when X happens, do Y.” It’s powerful but rigid. A thermostat is automation: when the temperature drops below 68°F, turn on the heat. Automation doesn’t learn, adapt, or handle surprises. It just follows the rules. For a deeper comparison, see our AI vs. automation explainer. Contractor example: Automatically sending a review request email 3 days after job completion. The email always goes out. It doesn’t decide whether THIS customer should get a different message.

Automation vs. AI

The key distinction: automation follows rules you set. AI learns patterns and makes judgment calls. Automation says “send an email at 9 AM every Tuesday.” AI says “this customer is most likely to respond to emails sent on Wednesday mornings based on their past engagement patterns — send it then.” Most contractor tools use a mix of both. The automation handles the routine workflows. The AI handles the parts that require judgment or pattern recognition.

B

Bias

When an AI system produces skewed or unfair results because of problems in the data it was trained on. If a hiring AI was trained mostly on data from male applicants, it might unfairly rate female applicants lower — not because anyone told it to, but because it learned the wrong pattern. Bias isn’t intentional, but it’s real, and it matters when AI makes decisions that affect people. Contractor example: An AI lead scoring system trained mostly on data from affluent neighborhoods might unfairly deprioritize leads from other areas, causing you to miss good customers.

Bot

A software program that performs automated tasks. A chatbot answers customer questions on your website. A social media bot auto-posts content. The term is broad — some bots are simple scripts, others use sophisticated AI. When someone says “bot” in a contractor context, they usually mean a chatbot on a website or an AI phone answering system. Contractor example: The chat widget on your website that answers “What areas do you serve?” at 11 PM without a human being awake.

C

Chatbot

An AI program designed to have text conversations with people. Modern chatbots use natural language processing to understand what someone is asking and generate helpful responses. They range from simple (picking from pre-written answers) to sophisticated (understanding complex questions and generating custom responses). Contractor example: A website chatbot that can answer questions about your services, check your availability, and book an estimate appointment — all through a text conversation.

Cloud Computing

Running software and storing data on remote servers (in a data center somewhere) instead of on your own computer or local server. When you use QuickBooks Online instead of desktop QuickBooks, that’s cloud computing. Almost all AI tools for contractors are cloud-based — you access them through a browser or app, and the heavy computing happens on someone else’s servers. Contractor example: Your field service software that your office team and field crews both access from different locations — the data lives in the cloud, not on any one computer.

Computer Vision

AI that can “see” and understand images and video. It’s the technology behind AI that measures roofs from satellite images, reads receipts from photos, identifies safety hazards on job sites, or recognizes defects in materials. Computer vision converts visual information into data that software can process. Contractor example: AI that analyzes a photo of water damage and identifies the likely source, extent, and repair scope — or that measures a roof from a drone photo for an estimate.

D

Data Privacy

The rules and practices around how personal and business data is collected, stored, used, and shared. When you use AI tools, your data (customer information, job costs, financial details) goes into those systems. Data privacy determines who can access that data, what they can do with it, and how it’s protected. For contractors, this matters most for customer data and financial information. See our AI safety and privacy guide for a deeper dive. Contractor example: When your AI CRM stores customer names, addresses, and payment information, data privacy policies determine whether the AI vendor can use that data for other purposes or share it with third parties.

Deep Learning

A type of machine learning that uses neural networks with many layers to learn complex patterns. “Deep” refers to the depth of the network (many layers), not philosophical depth. Deep learning powers the most impressive AI capabilities — natural language understanding, image recognition, voice synthesis. You don’t need to understand how it works. You just need to know that when someone says “deep learning,” they mean “really sophisticated pattern recognition.” Contractor example: The AI behind voice phone answering systems uses deep learning to understand accents, background noise, and conversational context — not just individual words.

E

Edge Computing

Processing data on the device itself (the “edge” of the network) instead of sending it to the cloud. This matters when you need fast responses or don’t have reliable internet. Instead of sending a photo to a server for analysis and waiting for the response, edge computing processes it right on your phone or tablet. Contractor example: An AI safety monitoring camera on a job site that detects someone without a hard hat and triggers an alert instantly — without needing to upload the video to a cloud server first. Works even if the job site has spotty internet.

Encryption

Scrambling data so that only authorized people can read it. When data is encrypted, it looks like gibberish to anyone who intercepts it without the encryption key. Most AI tools use encryption for data in transit (while it’s being sent over the internet) and data at rest (while it’s stored on servers). Contractor example: When your AI bookkeeping tool syncs with your bank account, encryption ensures that your financial data can’t be intercepted and read by hackers during the transfer.

F

Fine-Tuning

Taking a pre-built AI model and training it further on specific data to make it better at a particular task. The base model understands language or images in general; fine-tuning teaches it the specifics of your domain. It’s like hiring an experienced tradesperson who knows construction and then training them on your company’s specific processes. Contractor example: A general-purpose AI writing tool that’s been fine-tuned on construction proposals, so it understands trade terminology, estimate formatting, and contractor-to-customer communication style.

G

Generative AI

AI that creates new content — text, images, audio, video, or code — based on patterns it learned during training. ChatGPT generates text. DALL-E generates images. These tools don’t copy from a database; they generate original content that’s new but follows the patterns they’ve learned. Contractor example: Using generative AI to draft a customer proposal. The AI doesn’t copy an old proposal — it generates a new one based on your job details, using patterns it learned from millions of business documents.

GPT (Generative Pre-trained Transformer)

A specific type of large language model created by OpenAI. GPT is the technology behind ChatGPT. “Generative” means it creates content. “Pre-trained” means it was trained on a massive dataset before you use it. “Transformer” refers to the underlying architecture. GPT has become almost a generic term for powerful AI language models, similar to how “Kleenex” became a generic term for tissues. Contractor example: When you ask ChatGPT to “write a follow-up email for a customer who received an estimate for a kitchen remodel three days ago,” it uses GPT to generate that email.

H

Hallucination

When an AI generates information that sounds confident and plausible but is completely made up. AI doesn’t “know” things the way humans do — it predicts what text should come next based on patterns. Sometimes those predictions are wrong, and the AI states false information with total confidence. This is one of the most important limitations to understand. Contractor example: You ask an AI to research local building codes for deck railings, and it confidently gives you a 36-inch height requirement — but your jurisdiction actually requires 42 inches. The AI didn’t look up the code; it generated what seemed likely. Always verify AI outputs against authoritative sources.

I

Inference

When an AI model makes a prediction or generates output based on new input. Training is when the AI learns. Inference is when it applies what it learned. Every time you ask ChatGPT a question or an AI tool categorizes a receipt, that’s inference — the model is using its training to handle new data. Contractor example: Your AI bookkeeping tool sees a charge of $423.17 from Ferguson Enterprises and infers it’s a plumbing supply purchase for Job #2847 — that’s inference based on training from your previous transactions.

IoT (Internet of Things)

Physical devices that connect to the internet and share data. Smart thermostats, connected security cameras, equipment sensors, GPS trackers on trucks — anything that’s a physical device sending data over the internet is IoT. AI processes the data these devices generate to provide insights and automation. Contractor example: GPS trackers on your trucks that feed location data to your AI scheduling system, which uses it to optimize dispatch by knowing exactly where each crew is in real time.

L

LLM (Large Language Model)

An AI model trained on massive amounts of text data that can understand and generate human language. “Large” refers to both the training data (billions of documents) and the model size (billions of parameters). ChatGPT, Claude, Gemini, and Llama are all LLMs. They’re the technology behind AI writing assistants, chatbots, and voice AI systems. Contractor example: The AI that powers your phone answering system is likely an LLM — it understands what callers say, generates appropriate responses, and handles the conversation naturally because it learned from billions of examples of human conversation.

M

Machine Learning (ML)

A type of AI where the system learns from data instead of being explicitly programmed. Instead of writing rules for every scenario, you feed the system examples and it figures out the patterns. Machine learning gets better over time as it processes more data — that’s the “learning” part. Contractor example: An AI estimating tool that starts with industry-average labor rates but learns YOUR actual labor rates from completed jobs. After 50 jobs, it estimates based on how YOUR crew performs, not generic averages.

Model

The trained AI system itself — the thing that takes input and produces output. When people say “AI model,” they mean the software that’s been trained and is ready to use. Different models are trained for different tasks: language models for text, vision models for images, prediction models for forecasting. Contractor example: The AI model behind your receipt scanning app was trained on millions of receipts, so it knows how to extract vendor names, amounts, dates, and line items from the messy photos you snap in your truck.

N

Neural Network

An AI system loosely inspired by the human brain, made up of interconnected nodes (“neurons”) organized in layers. Data flows through these layers, and the connections between nodes are adjusted during training to produce accurate outputs. You don’t need to understand the technical details — just know that neural networks are the underlying architecture behind most modern AI. Contractor example: The computer vision system that measures roofs from satellite images uses a neural network trained on thousands of roof images to distinguish roof edges from shadows, trees, and neighboring structures.

NLP (Natural Language Processing)

AI’s ability to understand and work with human language — reading text, understanding speech, generating responses. NLP is what makes AI chatbots, voice assistants, and document processing possible. It handles the messiness of real human communication: slang, typos, accents, incomplete sentences, context-dependent meaning. Contractor example: When a customer texts “hey my heat aint workin again its freezing” and your AI system understands this is an HVAC emergency service request — that’s NLP handling informal, ungrammatical language and extracting the right meaning.

O

OCR (Optical Character Recognition)

Technology that reads text from images — converting a photo of a document into actual text data that software can process. OCR is the first step in receipt scanning, invoice processing, and document digitization. Modern OCR powered by AI is dramatically more accurate than older versions, handling handwriting, low-quality photos, and damaged documents. Contractor example: Snapping a photo of a handwritten material list from the job site and having AI convert it into a digital purchase order with item names, quantities, and supplier information.

P

Predictive Analytics

Using historical data and AI to forecast future outcomes. Instead of looking at what happened (descriptive analytics), predictive analytics tells you what’s likely to happen. It’s pattern recognition applied to business decisions. Contractor example: AI that analyzes your past 3 years of booking data and predicts that you’ll hit capacity in week 3 of May — giving you time to line up a subcontractor or start a hiring push before you’re overwhelmed.

Prompt

The input you give to an AI system — the question, instruction, or request that tells the AI what you want. The quality of your prompt directly affects the quality of the output. A vague prompt gets a vague answer. A specific, detailed prompt gets a specific, useful answer. Prompting is a skill, but it’s more like “giving clear instructions to a new employee” than anything technical. Contractor example: Instead of prompting “write a proposal,” prompting “write a proposal for a 200-square-foot deck build for a residential client, using composite decking, including a 12-month warranty, with a professional but friendly tone” gets dramatically better results.

R

RAG (Retrieval-Augmented Generation)

A technique that improves AI accuracy by giving it access to specific, trusted information before it generates a response. Instead of relying only on what it learned during training, the AI retrieves relevant documents or data and uses that information to inform its answer. This reduces hallucinations and makes outputs more accurate and current. Contractor example: An AI assistant for your company that has access to your price list, service agreements, and warranty terms. When a customer asks about warranty coverage, the AI retrieves your actual warranty document and answers based on YOUR policy — not generic information.

S

SaaS (Software as a Service)

Software you access through the internet and pay for with a monthly or annual subscription, instead of buying and installing it on your computer. Almost every AI tool for contractors is SaaS — you sign up, pay monthly, and use it through a browser or app. You don’t own the software; you rent access. Contractor example: ServiceTitan, Jobber, Buildertrend, QuickBooks Online — all SaaS products. You pay monthly, access them from anywhere, and the company handles updates and maintenance.

Sentiment Analysis

AI’s ability to determine the emotional tone of text — positive, negative, or neutral. It reads customer reviews, emails, or social media posts and tells you how people feel about your business. Contractor example: AI that scans your Google reviews and identifies that 3 of your last 10 reviews mention “cleanup” negatively — a specific, actionable insight you might miss reading reviews individually.

Supervised Learning

A type of machine learning where the AI is trained on labeled examples — data that already has the correct answers attached. The AI learns to match inputs to outputs by studying these examples. It’s like training an apprentice by showing them completed work and explaining what’s right and why. Contractor example: Training an AI expense categorizer by showing it 1,000 receipts that you’ve already labeled “materials,” “fuel,” “equipment rental,” etc. The AI learns the patterns and starts categorizing new receipts on its own.

T

Token

A chunk of text that AI processes as a single unit. A token isn’t always a whole word — it might be a word, part of a word, or a punctuation mark. AI models have limits on how many tokens they can process at once. This matters when you’re working with long documents or detailed prompts. Contractor example: If you try to paste a 50-page specification document into ChatGPT and it says it’s too long, you’ve exceeded the token limit. Break it into smaller sections.

Training Data

The information used to teach an AI model. The quality and breadth of training data directly determines how well the AI performs. An AI trained on millions of construction documents will understand contractor language better than one trained only on general internet text. Contractor example: An AI estimating tool trained on 500,000 real construction estimates will produce better estimates than one trained on 5,000 — because it’s seen more variations, more trade-specific patterns, and more real-world pricing data.

Transfer Learning

Taking an AI model trained for one task and adapting it for a related task. Instead of training from scratch, you start with a model that already understands the basics and teach it the specifics. This is faster and requires less data than building a model from zero. Contractor example: A general-purpose image recognition model that’s been adapted (transfer learned) to identify specific roofing materials, damage types, or safety violations on construction sites.

U

Unsupervised Learning

A type of machine learning where the AI finds patterns in data without being told what to look for. No labels, no correct answers — just data. The AI discovers groupings, trends, and anomalies on its own. Contractor example: AI that analyzes your customer database and discovers distinct groups you didn’t know existed — like “homeowners who always request the cheapest option” vs. “homeowners who always upgrade to premium materials.” You didn’t tell it to find these groups; it discovered the pattern.

V

Voice AI

AI that can understand spoken language (speech-to-text), generate spoken responses (text-to-speech), and manage voice conversations. Voice AI combines several technologies: speech recognition, natural language processing, and voice synthesis. Modern voice AI sounds remarkably human — natural cadence, appropriate pauses, and conversational tone. Contractor example: AI phone answering systems that have full conversations with callers — understanding their questions, asking follow-ups, booking appointments, and sounding like a natural human receptionist. See our AI answering services guide for tool comparisons.

W

Workflow Automation

Connecting multiple automated steps into a complete process. Individual automations do one thing; workflow automation chains them together. It’s the difference between a single power tool and an assembly line. Contractor example: A complete lead-to-customer workflow: AI answers the phone → captures lead info → creates CRM record → sends confirmation text → schedules estimate appointment → assigns to estimator → sends reminder 24 hours before → sends follow-up after estimate → triggers review request after job completion. Each step is automated; the workflow connects them all.

Terms You'll Hear But Don't Need to Worry About

The AI world loves jargon. Here are some terms you might encounter that sound important but mostly don’t affect how you use AI tools as a contractor:

• Attention mechanism — The technical architecture inside transformer models. It’s why AI can understand context. You don’t need to know how it works any more than you need to understand combustion chemistry to drive a truck.
• Backpropagation — How neural networks learn from mistakes. It’s the math behind training. Let the engineers worry about it.
• Embedding — How AI represents text or data as numbers internally. Important for developers, irrelevant for users.
• Epoch — One complete pass through the training data during model training. Unless you’re building your own AI model (you’re not), this doesn’t matter.
• Gradient descent — The mathematical process AI uses to improve during training. File this under “interesting but not useful for choosing a CRM.”
• Hyperparameter — Settings that control how a model trains. The AI equivalent of factory settings on equipment — the manufacturer handles it.
• Latent space — The internal representation AI uses to understand relationships between concepts. Fascinating for researchers. Irrelevant for contractors.

If a vendor is using these terms in a sales pitch, they’re either showing off or trying to confuse you. The tools that matter explain what they DO, not how they work internally. For more on cutting through AI hype, see our AI myths vs. reality guide.

How to Use This Glossary

Bookmark this page. When you’re reading about an AI tool, evaluating a vendor pitch, or trying to understand an article about AI in construction, come back here for definitions that make sense.

The most important terms for contractors evaluating AI tools are:

• AI vs. Automation — Know the difference so you know what you’re buying
• SaaS — Understand the subscription model for contractor tools
• API — Know that this is how tools connect to each other
• Hallucination — Understand that AI can be wrong, so always verify critical information
• Data Privacy — Know what happens to your data when you use AI tools
• OCR — The technology behind receipt scanning and document processing
• NLP / Voice AI — The tech behind AI phone answering and chatbots
• Predictive Analytics — How AI forecasts demand, capacity, and costs

You don’t need to understand every term on this page. You need to understand enough to make informed decisions about which AI tools are right for your business and to see through vendor hype. If you’re ready to start evaluating tools, our tool selection guide puts this knowledge into practice.