Resources · Glossary
Plain words for the jargon.
Every term we use on the site, explained the way we’d explain it to a colleague — no computer-science degree required.
AI & data
- AI model
- The “brain” you ask questions to — Claude, ChatGPT, Gemini or Copilot. It writes the answer, but on its own it knows nothing about your business; it has to be connected to your data safely.
- MCP (Model Context Protocol)
- A standard “plug” that lets any AI model connect to a data source in a controlled way. Because it’s a shared standard, you can switch from one AI model to another without rebuilding the connection. AccuraSee Gateway speaks MCP.
- AccuraSee Gateway
- The secure doorway between AI models and your data. Nothing reaches your data except through it — and it checks who is asking on every single question.
- AccuraSee Prism
- A governed home for your own AI apps, saved views and dashboards, so people build on company data inside the rules instead of in random tools.
- Semantic layer
- The single agreed definition of each number — for example, exactly what counts as “revenue” or “vacancy”. It means the AI’s answer and your reports always use the same definition, so you never get two versions of the truth.
- Unified data model (CDM)
- One tidy, combined picture of data that normally lives in many separate systems. Instead of pulling numbers from five places by hand, everything is brought together once.
- Governed views
- Pre-approved, read-only “windows” onto your data. The AI can look through these windows but can’t touch the underlying database or run anything risky.
- The spine
- Our shorthand for that one governed data foundation. You build the spine once; the AI models, dashboards and apps on top can all change around it without rebuilding it.
- Source citation
- Every answer shows where its numbers came from, so you can check the working — like footnotes on a report.
- Golden set
- A fixed list of questions with known-correct answers that the system is tested against continuously, to prove it stays accurate over time.
Security & trust
- Identity-aware
- The system always knows which real person is asking, so it can apply that person’s permissions to every answer.
- Entra ID / OAuth 2.1
- Microsoft’s standard sign-in (Entra ID) and the secure, industry-standard way apps prove who you are (OAuth 2.1). It’s the same kind of login your company already trusts.
- Row-level security (RLS)
- Rules that decide which rows of data each person is allowed to see. Two people can ask the same question and correctly get different answers, based on what each is cleared for.
- Audit trail
- A tamper-resistant log of who asked what, when, and the exact query that ran — so every answer is accountable and reviewable after the fact.
- Zero-retention
- The AI model processes your question and then keeps nothing. Your data isn’t stored by the model afterwards, and isn’t used to train it.
- In your tenant
- A “tenant” is your own walled-off space in the cloud. Running in your tenant means the data and the system live in your environment, under your control — not on someone else’s platform.
Finance & industry terms
- BI (Business Intelligence)
- Turning raw data into clear reports and dashboards that help you make decisions.
- Contract logic
- The billing rules a contract sets — different prices for different services, caps, price ladders, retainers — applied automatically so every invoice comes out right.
- Group / koncern reporting
- Combining the numbers from all the companies in a group into one consolidated view.
- NOI (Net Operating Income)
- A property’s income after its running costs — a core measure of profitability in real estate.
- ÄTA
- A Swedish construction term for changes and extra work ordered during a project (ändrings-, tilläggs- och avgående arbeten). It has to be tracked and billed, or money quietly leaks.
- Kalkyl
- The structured cost budget for a project — what each cost type is expected to cost. Comparing the kalkyl against actuals shows where a project is drifting.
Ask your business data anything. Get an answer in seconds.
Start with a discovery call — we'll scope a low-risk pre-study and show you what's answerable on your data.
What to expect: a 30-minute call · we map your data sources · a live answer on your data — no pressure.