AI Glossary & Jargon Buster
Reference
Plain-English definitions of every AI term you will encounter — with property management examples.
Core Concepts
| Term | What it means | Property example |
|---|---|---|
| LLM | Large Language Model — an AI trained on vast amounts of text that can understand and generate human language | Claude, ChatGPT, Gemini are all LLMs |
| Prompt | The instruction or question you type to the AI | “Draft a rent arrears notice for a tenant 14 days overdue” |
| Context window | The amount of text an AI can “see” and remember in one conversation — measured in tokens | A 10-page lease fits easily; a 200-page portfolio report may need to be split |
| Hallucination | When an AI confidently states something that is false or made up | AI inventing a clause that isn’t in the lease, or giving wrong market rent figures |
| Token | A chunk of text (roughly 0.75 words). AI models have limits on how many tokens they can process at once | 1,000 tokens ≈ 750 words ≈ about 1.5 pages of text |
| Temperature | A setting that controls how creative vs predictable AI responses are. Low = consistent; high = more varied | Use low temperature for formal notices; higher for brainstorming |
Tools & Features
| Term | What it means |
|---|---|
| Project / Custom GPT | A saved workspace with permanent instructions and uploaded files — your AI standard operating procedure |
| System prompt | Background instructions that shape every response in a Project — your firm’s rules, tone, and persona baked in |
| Skill / Agent Skill | A reusable set of instructions you paste in to give the AI a specific capability (e.g. Negotiation Coach) |
| Artifacts | In Claude: files (Word docs, spreadsheets) that appear in a separate panel, ready to download |
| Customise | Personal settings that apply to every conversation — your name, role, firm, and preferences |
Common Misconceptions
❌ “AI knows my portfolio” — It doesn’t. LLMs have no access to your specific data unless you paste it in or upload it.
❌ “If it sounds confident, it’s right” — LLMs can be confidently wrong. Always verify specific facts, figures, and legal advice.
❌ “It remembers last time” — Each new chat starts fresh. To continue a conversation, reopen the same chat thread.
✓ What to trust — Drafting, summarising, explaining, brainstorming. AI is excellent at communication tasks where you review before sending.
✓ What to verify — Specific legal clauses, rent figures, dates, compliance obligations. Treat as a first draft, not a final answer.
