AI and digital twin glossary
The words used across AccessMe and in these guides, each defined in a sentence or two. Where everyday usage and the technical meaning differ, both are given.
- Digital twin (of a person)
- A conversational AI that answers from one real person's own material, in that person's voice and point of view, labeled as an AI representation, with the real person able to read shared conversations and follow up. On AccessMe, every twin is an Advisor Twin or a Gatekeeper Twin.
- Advisor Twin
- A twin that answers its creator's audience from the creator's material and point of view. It keeps the creator reachable. Available on every AccessMe plan.
- Gatekeeper Twin
- A twin that interviews the people who want its creator's time, scores each conversation against the creator's criteria, and shows the creator a ranked shortlist with summaries in their dashboard. It does not notify, book or decide. Available on every AccessMe plan.
- Panel
- A conversation in which several twins answer the same question. Each twin answers only from its own creator's material and never sees the other twins' answers; a built-in AccessMe moderator adds a summary. Each twin on a panel is one session at that twin's own price.
- Moderator
- The built-in AccessMe participant in a panel that summarizes the twins' separate answers and points out where they agree and differ. It does not answer on its own behalf.
- Creator / Original
- The real person a twin is trained on and represents. The AccessMe support area calls this person an Original; the rest of the site says creator. They mean the same thing.
- Audience
- The people who talk with a twin: readers, clients, students, members, founders, makers. An audience member can talk with any published twin and uses one credit balance across all of them.
- Training (everyday sense)
- Giving a twin your material and answering the guided questions about your voice. On AccessMe this does not retrain any AI model; see Context.
- Training (technical sense)
- The process of building a language model by adjusting its internal weights on large amounts of text. Permanent, shared by every user of the model, and done before release. Your material does not go through this on AccessMe.
- Context
- The material a language model can see while answering one question: the conversation so far plus any passages placed in front of it. Read for that answer and then discarded. A twin works by placing the relevant parts of its creator's material in context.
- Retrieval
- Finding the passages of a creator's material most relevant to a question so they can be placed in the model's context. What makes a twin's answers specific to the creator rather than generic.
- Large language model (LLM)
- An AI model trained to predict text, which makes it able to answer questions, write and converse. The general assistants (ChatGPT, Claude, Gemini) are LLMs; a twin uses an LLM anchored to one person's material.
- Hallucination
- A fluent, confident answer from a language model that is false: an invented quote, fact or citation. Reduced, not eliminated, by giving the model the right material in context.
- Memory (in AI)
- Stored information that a product re-supplies to a model as context so it can appear to remember. A model itself retains nothing between conversations. An AccessMe twin remembers returning audience members.
- Voice
- A person's written manner and point of view: formality, how they open and close, what they emphasize, what they would never say. A twin carries voice from the creator's material and the guided voice questions. Not an audio voice; twins are text chat only.
- Session
- One conversation with one twin, open for up to 24 hours, with a fair-use message limit. The unit a twin is free or paid in.
- Credit
- The unit audience members prepay with on AccessMe: $1, usable only on AccessMe, one balance for every twin including panels. Paid twins list their price in credits per session.
- Studio Credits
- The monthly allowance of credits included with a creator's or organization's plan, plus any extra bought at $1 each. A free twin's sessions use one Studio Credit each. Included monthly Studio Credits expire at the end of each billing period.
- Normal mode
- A conversation mode in which the creator can read the conversation and follow up, and the conversation contributes to audience insights. The audience member chooses the mode on paid plans.
- Private mode
- A conversation mode in which not even the creator can read the messages, and the conversation is excluded from audience insights. Chosen by the audience member, on paid plans.
- Audience insights
- On paid plans, a view for the creator of what people are talking with the twin about most, drawn from normal-mode conversations only.
- Ranked shortlist
- What a Gatekeeper Twin produces: the people it interviewed, ordered by how well each conversation matched the creator's criteria, each with a summary. Shown in the creator's dashboard; the creator decides who to meet.
- AI disclosure
- The statement on every twin page that the twin is an AI representation trained on its creator's content, not the person themselves. Cannot be turned off.
- Discover
- The public page at discover.accessme.ai where anyone can browse published twins and start a conversation or build a panel.
- Twin page
- A twin's own page at name.accessme.ai, where the audience chats with it from any device with no app to install.
- Answer engine optimization (AEO)
- Writing and structuring content so that AI assistants and answer engines can find, quote and attribute it when people ask them questions. The reason this Learn area answers questions directly and defines its terms.
Ready to build yours?
Every plan, including Free, includes an Advisor or Gatekeeper twin trained on your own material. Your content stays yours.