# Creation ## Description The Creation workflow lets you build a brand‑new agent from scratch. In a single guided flow you name the agent, set its personality, choose a language model, upload knowledge resources, and generate an embed code—so you can deploy a customized AI tutor anywhere learners need it. ![](/images/creation.png) --- ## Target Audience **Administrator** --- ## Features #### Create Agent Wizard A streamlined interface (Settings ➜ Prompts ➜ Tools ➜ Datasets ➜ Embed) that takes you step by step from concept to launch. #### Rich Settings Panel Upload profile images, set visibility (Superadministrator‑only or User‑facing), pick a base LLM (OpenAI or Gemini), and categorize the agent for easy search. #### Prompt Configuration Define a **System Prompt** (agent persona), **Proactive Greeting**, and **Guided Prompts** that appear as starter questions. #### Tool Toggles & Safety Controls Enable features like **Web Search** or **Screen Share**, and specify how the agent should respond to inappropriate queries. #### Dataset Uploads Add **PDFs**, **URLs**, **Google Drive**, **Dropbox**, or **OneDrive** files to form the agent’s knowledge base. #### Conversation History Analytics Review every chat, export transcripts, view sentiment, and filter by date, topic, or category. #### Embeddable Widget Generator Produce a code snippet with options for anonymous or trackable usage, context awareness, disclaimers, and launcher icon customization. --- ## How to Use (step by step) #### Start the Creation Process - Log in and click **Create** - The **Create Agent** screen opens on the **Settings** tab #### Fill Out the Settings Tab - **Profile Image** – upload an avatar for the agent - **Agent Name** – e.g., “Bio AI” - **Description** – short phrase like “Biology tutor that assists learners with Intro to Biology” - **Category** – choose relevant tags (biology, science, learning) **Agent Visibility –** - **Superadministrators** → only staff can see/use (assessments, admissions, internal data) - **Users** → everyone can access (general tutoring) - **Base Model** – select OpenAI default or Gemini #### Define Prompts - Open the **Prompts** section - **System Prompt** – outline persona (e.g., Socratic biology tutor) - **Proactive Message** – greeting shown before a learner’s first message - **Guided Prompts** – AI‑suggested starter questions #### Configure Additional Settings - **LLM Selection** – choose the specific model version - **Tools** – toggle extras (Web Search, Code Interpreter, Image Generation, etc.) - **Safety** – set rules for inappropriate content and define the agent’s fallback response #### Review the History Tab (post‑launch) Inspect all conversations, export chats, and filter by sentiment, date, or topic. #### Upload Knowledge in Datasets Add files, folders, or cloud links. These resources train the agent’s retrieval system. #### Embed the Agent Elsewhere - Open the **Embed** tab - Choose **Default** or **Advanced** view - Decide on **Anonymous vs. Trackable** usage - Enter the target **Website URL** - Optional toggles: **Context Aware**, **Safety Disclaimer**, **Open by Default**, **Icon** - Click **Create Embed** and copy the snippet #### Go Live Deploy the embed code or share the agent link. Learners (or superadministrators) can now chat with the agent immediately. --- ## Pedagogical Use Cases #### Course‑Specific Tutors Build agents like “Bio AI” or “Macro Econ Coach” that draw on syllabus PDFs and lecture slides. #### Admissions & Advising Assistants Create Superadministrator‑visible agents to handle FAQs, application guidance, or scholarship queries. #### Assessment Authoring Helpers Generate agents restricted to administrators that draft quiz questions or grading rubrics. #### Research Support Bots Upload journal articles and datasets so the agent can summarize findings and suggest next steps for user projects. #### Campus‑Wide FAQ Bots Embed a general “Campus Help Desk” agent across the LMS to answer policy, tech, or financial‑aid questions 24/7. --- Once you complete these steps, your agent is fully configured—ready to guide learners or staff, depending on the visibility you selected.