Glossary
A reference guide to the terminology used throughout EVE Builder.
Account library
Section titled “Account library”A shared collection of models made available to all users within an account, without requiring individual sharing. Models are added to the account library by being promoted by their owner and removed by being demoted. Appears as the Account Library Models section on the EVE Builder home. See Models.
AI Only
Section titled “AI Only”A conversation mode where the AI drives all probing and prompting based solely on the context and topics you provide. No predefined model is required. Best suited to fast setup, exploratory research, and multilingual studies. See AI Only.
AI probe
Section titled “AI probe”A probe generated by the AI when a participant’s response is relevant but doesn’t match a keyword defined in the linked model. Available in Guided AI conversations as a safety net alongside Model Probes. The maximum number of AI Probes is configurable in the conversation flow settings. See Guided AI.
Avatar
Section titled “Avatar”The image displayed to participants alongside the Conversation Companion during a conversation. Configurable per conversation — defaults to the Eve avatar if not set. See Conversation settings.
Chained EVE
Section titled “Chained EVE”An alternative name for a Discussion — two or more published conversations linked together into a single extended session. The terms are used interchangeably within EVE Builder. See Discussions.
Concise question wording
Section titled “Concise question wording”An advanced setting that removes affirmations and paraphrasing from EVE’s probes and prompts. When enabled, EVE responds with concise, direct single sentences rather than longer conversational phrasing. See Conversation settings.
Content ID
Section titled “Content ID”The unique identifier (GUID) for a published conversation. Used to link a conversation to a Survey Shaper survey via the Eve Source field. Visible in the content table by enabling the Show Content ID toggle, and also displayed in the Publish Success dialog. Also referred to as the EVE Builder Project ID. See Publish and deploy.
Conversation
Section titled “Conversation”A single AI-moderated feedback session configured in EVE Builder. Defines the conversation type, introduction, probes, prompts, flow rules, and companion settings. Must be published before it can be deployed. See Getting started.
Conversation companion
Section titled “Conversation companion”The AI persona presented to participants during a conversation. Has a configurable name (default: Eve) and avatar image. The companion name and avatar are set per conversation in the Preview step. See Conversation settings.
Conversation outro
Section titled “Conversation outro”The closing message EVE delivers at the end of a conversation. Configured in the Conversation Flow step alongside the introduction and final question. See Conversation settings.
Danger words
Section titled “Danger words”A safety feature available in Guided AI and Model Only conversations. When enabled, EVE monitors participant responses for specified words or phrases and responds with a configured message when one is detected. Supports a default danger words list, a custom list defined in the model, or both. Response fields include pre-filled example text with [Insert company name] placeholders that must be updated before publishing. See Conversation settings.
Discussion
Section titled “Discussion”A project type that links two or more existing published conversations together into a single extended session. Also referred to as Chained EVE. EVE works through each component conversation in sequence. Supports optional randomisation of conversation order and an optional closing question. See Discussions.
See Save draft.
EVE Builder Project ID
Section titled “EVE Builder Project ID”See Content ID.
Final question
Section titled “Final question”An optional static question asked by EVE at the end of every conversation, before the outro. Enabled via the Ask Static, Final Question checkbox in the Conversation Flow step (enabled by default). Not counted toward the Maximum follow-up questions limit. See Conversation settings.
Fixed (language rule)
Section titled “Fixed (language rule)”A language rule setting that instructs EVE to use a specific language throughout the conversation, regardless of what language the participant writes in. One of two Language Rule options alongside Native (Respondent Language). See Conversation settings.
Guided AI
Section titled “Guided AI”A conversation mode where a predefined model structures probing and prompting, with the AI acting as a safety net for responses that fall outside the model. Combines the precision of model-led conversations with AI flexibility. See Guided AI.
Keyword
Section titled “Keyword”A word or phrase defined within a Sub Topic in a model. When a participant mentions a keyword in their response, EVE probes on the parent sub-topic. Each keyword has a Tagging rule that controls how it is matched against participant responses. Keywords are the core matching mechanism in Guided AI and Model Only conversations. See Models.
Language rule
Section titled “Language rule”An advanced setting that enables EVE to converse in languages other than English. Two options: Native (Respondent Language) — EVE responds in whatever language the participant uses; Fixed — EVE uses a specific language you select. See Conversation settings.
Maximum AI probes
Section titled “Maximum AI probes”A conversation flow setting in Guided AI conversations that caps the number of AI-generated probes EVE can ask. Separate from the Maximum Model Probes limit. Default: 1. See Guided AI.
Maximum follow-up questions
Section titled “Maximum follow-up questions”A conversation flow setting that caps the total number of probes and prompts EVE will ask a participant across a conversation. The static Final Question does not count toward this limit. Default: 2. See Guided AI and Model Only.
Maximum model probes
Section titled “Maximum model probes”A conversation flow setting that caps the number of keyword-triggered probes from the linked model that EVE can ask. Default: 1. See Guided AI and Model Only.
Maximum random prompts
Section titled “Maximum random prompts”A conversation flow setting that caps the number of prompts EVE can deliver on topics the participant hasn’t mentioned. Set to 0 to disable random prompting entirely. Default: 1. See Guided AI and Model Only.
A reusable framework of Topics, Sub Topics, and Keywords that structures probing and classifies responses in Guided AI and Model Only conversations. Models also carry the probe and prompt wording EVE uses when keywords are triggered. Can be shared across conversations and sub-accounts. See Models.
Model editor
Section titled “Model editor”The interface within EVE Builder where you build and edit the structure of a model — adding topics, sub-topics, keywords, and probe/prompt wording. Accessible when configuring a Guided AI or Model Only conversation, or directly from the Models section. See Models.
Model Only
Section titled “Model Only”A conversation mode driven entirely by a predefined model, with no AI-generated responses. EVE follows the model’s rules exactly. Best suited to scenarios requiring complete control, consistency, and sensitive topics. See Model Only.
Model probe
Section titled “Model probe”A probe triggered when a participant’s response contains a keyword defined in the linked model. Distinguished from an AI Probe in Guided AI conversations, where both types can occur. See Models and Guided AI.
Native (respondent language)
Section titled “Native (respondent language)”A language rule setting that instructs EVE to detect the language the participant uses and respond in kind throughout the conversation. One of two Language Rule options alongside Fixed. See Conversation settings.
A follow-up question EVE asks when a participant mentions a keyword from the linked model (a Model Probe) or something relevant detected by the AI (an AI Probe). Probe wording is configured in the Model Editor per sub-topic. See Models.
Probe on terms in initial response only
Section titled “Probe on terms in initial response only”A conversation flow setting that, when enabled, restricts EVE to probing only on keywords mentioned in the participant’s very first response. When disabled, EVE can probe on keywords mentioned at any point in the conversation. See Guided AI and Model Only.
Prompt
Section titled “Prompt”A question EVE asks to introduce a topic the participant hasn’t mentioned. Prompt wording is configured in the Model Editor per sub-topic. The frequency of prompting is controlled by the Maximum Random Prompts setting. See Models.
Published
Section titled “Published”A conversation that has been published is available for deployment via Survey Shaper. Published status is shown in the content table. If a published conversation is subsequently edited, it must be republished for the changes to take effect. See Publish and deploy.
Randomness
Section titled “Randomness”An advanced setting that controls how creatively EVE phrases its probes and prompts. Ranges from 0 to 1 — higher values produce more varied wording between participants; lower values produce more consistent phrasing. Default: 0.7. See Conversation settings.
Save draft
Section titled “Save draft”Saves the current conversation configuration without publishing. The conversation remains Unpublished and cannot be deployed until published. Available throughout the configuration stepper. See Publish and deploy.
Sub topic
Section titled “Sub topic”A specific aspect within a Topic in a model (e.g. Topic: “Customer Service”, Sub Topic: “Wait Times”). Each sub-topic contains Keywords that trigger probing, and carries its own probe and prompt wording. See Models.
Tagging rule
Section titled “Tagging rule”Controls how EVE matches a keyword against participant responses. Each keyword has one tagging rule assigned. Options include Exact word or phrase, Starting with, Ending with, Containing, Any of these words, All of these words, in any order, All of these words, in this order, in a sentence, Do not contain, and Custom (regular expression). See Models.
A broad category of interest defined in a model (e.g. “Price”, “Customer Service”). Topics contain Sub Topics, which in turn contain Keywords. In AI Only conversations, topics are generated from context and used by the AI to guide the overall scope of the conversation — they don’t contain keywords or sub-topics. See Models.
Unpublished
Section titled “Unpublished”A conversation that has not yet been published, or has been edited since its last publish. Unpublished conversations cannot be deployed via Survey Shaper. See Publish and deploy.