Working with Minikai, you may encounter terms specific to our platform or the tech and AI world. Some are internal, others from broader software or automation. This worksheet explains them in plain language to help you understand quickly—no technical background needed.
Term | Meaning |
Mini | A Mini is Minikai’s AI assistant that helps you create, organise, and summarise information about the person under care. You can give it instructions (spoken or typed) to draft progress notes, prepare reports, pull details from existing records, or answer questions based on the information your organisation has stored. Minis are designed to save time on administrative tasks and support more consistent, person-centred care. |
Prompt | A typed or spoken instruction given to a Mini. This is how you “ask” the Mini to do something, like summarise notes, draft a plan, or find information. |
Record | A document like a plan, note, form, or assessment that a Mini can help generate, search, or summarise. This includes things like incident reports, behaviour support plans, mealtime plans, or progress notes. Minis draw on data sources to help answer user questions. Understanding which data sources are important to you helps us improve the quality of Mini responses. |
Eval or Evals | Short for “evaluation”. These are internal tests we run to check whether a Mini’s response is accurate, helpful, and aligned with expected behaviour. |
Quality | Refers to how accurate, useful, and appropriate a Mini’s response is. Quality is measured through evals — structured reviews where Mini responses are checked against real examples or expected outcomes. These evaluations help ensure that content generated by the Mini aligns with professional standards, supports clinical or operational decision-making, and reflects the context of the person under care. Minikai regularly uses evals to monitor and improve the quality of Mini outputs over time. |
Preferences | Mini's store preferences to provide helpful, consistent context over time across interactions. While you will be able to create new preferences through conversation, an initial list helps us tailor Mini behaviour from the start. |
Context-Aware Response | The Mini’s ability to tailor its response based on the task you’re doing, your role, or the document you’re working on. It avoids “generic” replies and aims to be fit-for-purpose |
Speech to Text | A feature where a Mini listens to your voice and turns it into written words |
LLM (Large Language Model) | The underlying AI technology used by the Mini. It processes natural language (what you type or say) and generates a helpful response. You don't need to know how it works — but you may hear the term in discussions about AI or product development. |
Human-in-the-loop | A process where a human reviews AI-generated content before it’s used for quality assurance or compliance. At Minikai, this is often done during evals or training phases. |
Qualitative User Testing | A research method used by Minikai to understand how real users interact with the Mini. It involves observing how people use the tool in actual workflows (e.g. writing notes, preparing plans) and gathering feedback on what works well, what’s confusing, or what needs improvement. Unlike surveys or analytics, which measure broad trends, qualitative testing focuses on depth — looking closely at individual experiences to uncover insights that lead to better design and functionality. |
Mixpanel | An analytics tool Minikai uses to track how people use the Mini — such as what features they interact with, how often they use certain prompts, or where they might get stuck. This helps the team understand patterns in real usage (not just feedback), so they can improve the Mini based on what users are actually doing — not just what they say. Importantly, the data used is anonymised and focused on behaviour, not personal content. |
Structured output | A response from the Mini that follows a consistent format like headings in a support plan or sections in an incident report instead of just a block of plain text. |
Story Mapping | A collaborative method Minikai uses to understand how people want to use the Mini in their real-world workflows. During story mapping, we break down a task — like writing a support plan or updating case notes — into smaller steps from the user’s perspective. This helps ensure the Mini is designed to support the full journey, not just isolated actions. It’s especially useful during onboarding or pilot design, where we want to make sure the Mini fits how your team already works. |