# Large Language Model (LLM) > A large language model (LLM) is an AI system trained on large amounts of text that can read, summarize, and generate language. In construction, it can help parse RFPs, sub proposals, and specifications for preconstruction teams. - Stages: Preconstruction, Estimating, Business Development - Concepts: LLMs - Published: 2026-08-28 - Canonical: https://buildr.com/library/large-language-model ## Why it matters in construction Much of preconstruction lives in unstructured text: RFPs, addenda, sub proposals, meeting notes, owner email threads, specification sections, and closeout checklists. Historically, people read that material and entered the useful details into a spreadsheet or CRM. An LLM can identify patterns in that material. It may recognize that "excludes all cutting and patching" is an exclusion or that "10 weeks ARO" is a lead time. It can also miss important language, so teams need source links and review when the answer affects a bid. ## How it works 1. **Training.** Model builders train the system on a large body of text to predict the next token. Through that process, it learns grammar, facts, and patterns in language. 2. **Tokens in, tokens out.** The software splits a prompt into tokens, roughly word fragments. The model produces a response one token at a time using the text that came before it. 3. **No memory by default.** Each request starts fresh. Project information must be included in the prompt or retrieved from another system. 4. **Instructions shape output.** A system prompt sets the role and output format. This can make a general model behave like a proposal reader or scope-gap checker. 5. **Post-processing.** Tools can request structured output, validate it against a schema, and use ordinary code for calculations. An LLM is a reading and writing system. It is not a database, and it should not be the only system performing estimate calculations. Reliable tools use retrieval for project facts, source links for review, and normal code for math. ## Example in practice Consider a commercial GC receiving a 240-page RFP for a $38M higher-education renovation. The precon manager wants the bid form requirements, bonding thresholds, submission deadline, and any unusual insurance language before the go/no-go meeting on Thursday. An LLM-based tool produces a one-page summary: bid due at 2:00 PM on the 19th, a 5 percent bid bond, a $10M umbrella requirement, builder's risk carried by the owner, and a phasing requirement that keeps the library open during finals week. The precon manager checks the cited sections, then calls the two subs who can handle the phasing. ## Go deeper - [GPT Models for Contractors: A Plain-English Guide](/blog/gpt-models-for-contractors.md) - [The Pros and Cons of ChatGPT in Construction](/blog/chatgpt-construction.md) - [AI and Construction: The GC's No BS Guide to What Works in 2026](/blog/ai-construction-gc-guide-2026.md) ## How Buildr applies this Buildr's AI, Kit, uses language models to read the documents flowing through preconstruction and turn them into structured pipeline, estimating, and workforce data. See [Buildr Platform](/platform). ## Related terms - [Tokens](/library/tokens.md): Tokens are the small chunks of text that AI language models read and write, roughly three-quarters of a word each. They determine both the cost of a request and how much of a construction document fits in a model's context window. - [Context Window](/library/context-window.md): A context window is the amount of text, measured in tokens, that an AI language model can consider in one request. It limits how much of a construction spec book or proposal set the model can see at once. - [Hallucination](/library/hallucination.md): A hallucination occurs when an AI language model states something confidently that is untrue or unsupported by the source. For example, it might invent a unit price or an exclusion that does not appear in a subcontractor's proposal. - [Prompt Engineering](/library/prompt-engineering.md): Prompt engineering is the practice of writing instructions, examples, and context for an AI language model so it produces reliable output, such as telling it exactly how to classify exclusions in a construction subcontractor proposal. - [Reasoning Models](/library/reasoning-models.md): Reasoning models are AI language models designed to spend more effort on a problem before answering. They can be useful for multi-step construction judgments, such as reconciling a spec conflict across three divisions or scoring a go/no-go with competing criteria. ## Referenced by - [AI Data Privacy (Training on Your Data)](/library/ai-data-privacy.md): AI data privacy in construction is about whether a vendor or model provider uses your estimates, sub pricing, and project data to train models used by other companies, and which contractual and technical controls prevent that use. - [Computer Vision](/library/computer-vision.md): Computer vision is AI that interprets images and video. In construction, it can read drawings for takeoff, identify symbols and rooms on plans, compare jobsite photos with progress plans, and check PPE compliance in camera feeds. - [Fine-Tuning](/library/fine-tuning.md): Fine-tuning further trains an existing AI model on a specific set of examples so it learns a particular style, format, or task. A construction firm might use it to teach a model its conventions for classifying subcontractor scope items. - [Prompt Injection](/library/prompt-injection.md): Prompt injection is an attack in which instructions hidden in content an AI model reads, such as a subcontractor proposal or RFP attachment, alter the model's behavior. It is a major security risk for construction AI tools that process outside documents. - [Shadow AI](/library/shadow-ai.md): Shadow AI is the use of consumer AI tools by construction employees without company approval. For example, an estimator might paste a sub proposal or owner contract into a free chatbot, putting confidential project data outside the firm's security and contractual controls. ## FAQ ### Is an LLM the same thing as ChatGPT? ChatGPT is a product built on top of an LLM. The product includes the chat interface, memory, and tools around the model. Construction software vendors can use the same class of models behind their own interfaces. ### Does an LLM understand construction? It may recognize common construction terms, such as a Division 07 scope or substantial completion certificate. It does not know your subs, historical costs, or bid package unless the tool supplies that information in the prompt or through retrieval. ### Can an LLM do math on an estimate? Not reliably on its own. Language models predict text, and arithmetic is a weak spot. Good tools have the model extract quantities and prices, then hand the math to normal code.