AmplifyAI
AmplifyAI is Mercer’s multimodel generative AI access platform.
Before you begin: privacy and responsibility
Mercer’s privacy position: AmplifyAI is a Mercer-provided university environment. Information is not visible to other Mercer users unless you intentionally share it. However, access to a private platform does not mean every category of university or personal information is appropriate to enter.
Important privacy notice: AmplifyAI is not approved for FERPA-regulated student education records or HIPAA-protected health information (PHI). Do not enter, upload, paste, or share personally identifiable information, student records, health information, or other confidential institutional information in AmplifyAI.
Use only information you are authorized to access, and use the minimum information necessary. When information is needed for a legitimate task, de-identify it whenever possible. Remove names, student or employee IDs, dates of birth, contact information, account numbers, grades, disability or accommodation information, medical information, credentials, and other sensitive details.
Do not enter passwords, authentication codes, Social Security numbers, payment-card or bank-account information, government identification numbers, protected research data, personnel records, admissions records, disciplinary records, or confidential legal information.
- Verify important facts, quotations, citations, calculations, policies, and dates against authoritative sources.
- Do not treat an AI response as a final academic, employment, legal, medical, financial, admissions, grading, or personnel decision.
- Students must follow the AI directions established by each instructor and assignment.
- Faculty and staff remain responsible for accuracy, accessibility, copyright, professional judgment, and required disclosure.
- Review conversations, artifacts, templates, and assistants before sharing them.
- If you are unsure whether information may be used, do not enter it. Consult the appropriate Mercer office or supervisor before proceeding.
Quick start
- Sign in at mercer.cr8.io.
- Start a New Chat from the Conversations tab.
- Choose one of the available lower-cost models that fits the task.
- Attach a file or image when the task depends on a source, and wait for the green check mark.
- Adjust reasoning, temperature, or response length only when useful.
- State the task, context, source, desired output, and limits in the prompt.
- Verify the response before using or sharing it.
1. Find your way around
The left side of AmplifyAI has three main work areas:
- Conversations: Start and return to chats.
- Assistants and prompts: Open or build assistants and reusable prompt templates.
- Group and transfer tools: Work with assistant groups or import and export conversations.
2. Start a conversation and attach a file
Type directly in the message box for a text-only task. For work based on a document, spreadsheet, PDF, presentation, or image, select the attachment control near the message box. Choose the file and wait until its name appears with a green check mark before sending your prompt.
Write a grounded file prompt
Example: Using the attached syllabus as the only authoritative source, create a 300-word plain-language explanation for first-year students. Cite the relevant syllabus section after each answer. List questions the syllabus does not answer. Do not invent missing requirements.
- Name the file and explain how the model should use it.
- Say whether it is the only source or one of several sources.
- Ask for page, heading, section, or row references when possible.
- Tell the model to identify missing information rather than inventing it.
- For an image, ask what is visibly present before requesting interpretation.
3. Choose a model
Select the model name to open the model list. Mercer will provide a range of models, with emphasis on the lower-cost choices. Match the model to the task and compare results when the work is important.
| Indicator | What it means |
|---|---|
| Camera | The model can usually analyze uploaded images. This does not necessarily mean it can create images. |
| Green $ | Lower relative cost. Use these models for routine drafting, summaries, brainstorming, and many everyday tasks. |
| Gold or red $ | Higher relative cost. These models may be available for selected needs, but Mercer’s standard choices will emphasize lower-cost models. |
| Options icon | Opens settings or details that may vary by model. |
A practical model-choice rule
- Routine work: Start with a lower-cost general-purpose model.
- Files and images: Choose a model showing the appropriate file or camera capability.
- Current information: Use Web Search when available, open the cited sources, and check their dates.
- More difficult work: Compare the result with another available model when accuracy or quality matters.
4. Adjust model settings
Open the selected model’s settings to find options such as Custom Instructions, Temperature, Response Length, and Reasoning Effort. The exact controls may vary.
| Control | Recommended use |
|---|---|
| Temperature | Lower settings produce more focused, repeatable wording. Higher settings produce more variety and can help with brainstorming. |
| Response Length | Sets a general preference. A specific word or item limit in the prompt is more reliable. |
| Reasoning Effort | Low or Off is usually enough for simple work. Medium is a useful default. Higher reasoning may help with more complex analysis but can take longer. |
5. Write a prompt that works
A useful prompt usually includes five parts:
| Part | Question | Example |
|---|---|---|
| Task | What should the model do? | Create a study guide. |
| Context | What should it know? | This is for first-year biology students. |
| Source | What information should it use? | Use only the attached chapter. |
| Output | What should the answer look like? | Use five headings and a 10-item practice quiz. |
| Limits | What should it avoid or verify? | Do not invent terms; cite the chapter section. |
Prompt pattern: Using [source or context], do [task] for [audience]. Return [format and length]. Follow [constraints]. Identify uncertainty and do not invent missing information.
Treat the first response as a draft. Ask the model to explain a choice, show evidence, reorganize the answer, identify uncertainty, or compare alternatives.
6. Understand answers and artifacts
A normal answer appears in the conversation. When AmplifyAI creates a document, code project, graphic, HTML page, or similar item, it may also create an Artifact. Select the colored artifact card to open it beside the conversation.
Artifact actions
The available buttons depend on the artifact type and your permissions. Common actions include preview, save, copy, download, email, share, edit, and delete. Hover over an icon to read its current label before selecting it.
7. Improve speed and choose account settings
Select your profile circle to open account options such as theme, color scheme, sharing, data, user-cost information, Settings, feedback, and sign out. A High Contrast color scheme is available.
Turn off Smart Focused Messages when responses are slow
- Select your profile circle.
- Select Settings.
- Open Configurations and scroll to Features.
- Clear the Smart Focused Messages checkbox.
Preserve conversations you will need again
Open Settings → Conversation Storage and switch to cloud storage when you want conversations preserved beyond the default archive period. Stored conversation history and model memory are different. A new conversation does not automatically know what was discussed in an earlier conversation, so paste or attach the relevant context when needed.
8. Use prompt templates and assistants
Use a Prompt Template when you repeat the same kind of request but change a few details. Use an Assistant when you want a reusable helper with a defined purpose, instructions, and approved sources.
Prompt templates
- Give the template a clear name and description.
- Write the reusable prompt.
- Use double braces for details that will change, such as
{{audience}}or{{length}}. - Test the template with several examples before sharing it.
Assistants
- Give the assistant one clear purpose.
- Write instructions that explain what it should do and what it should not do.
- Upload only current, authoritative sources you are allowed to use.
- Test normal questions, unclear questions, missing information, and requests outside its purpose.
- Assign someone to review the sources and update or retire the assistant.
Layered Assistants and group tools
A Layered Assistant is a way to connect several specialized assistants. The first assistant reads the user’s question and sends it to the assistant best suited to answer it. For example, one entry point could direct a question to separate assistants for technology help, teaching support, or policy guidance.
Group tools allow people to work with a shared collection of assistants or move conversations into or out of AmplifyAI. Before using either feature, decide who owns the content, who may use it, who will keep it current, and what users should do when the system cannot answer.
9. Verify, save, and troubleshoot
Verification checklist
- Can I locate the claim in an authoritative source?
- Are citations and links real, relevant, and current?
- Do quotations match the source?
- Are calculations and totals correct?
- Did the model actually use the file or source I supplied?
- Does the result meet accessibility, privacy, copyright, and disclosure requirements?
Troubleshooting
| Problem | Try this |
|---|---|
| Slow response | Clear Smart Focused Messages, use Low or Off reasoning for simple work, shorten the prompt, or begin a new focused conversation. |
| File not recognized | Wait for the green check, mention the filename, reattach it, remove password protection, or convert scanned content to searchable text. |
| Weak or generic answer | Add context, a source, an audience, an output format, and clear limits. |
| Fabricated citation | Stop using the citation and locate the original source independently. |
| Missing feature | Check the selected model, settings, enabled tools, and your user role. |
10. Known limitations
11. Get help
Contact the Mercer IT Help Desk for sign-in, multifactor authentication, account activation, browser or device access, outages, and provisioning.
Contact Academic Technology at academictechnology@mercer.edu for model choice, prompts, files, assistants, templates, artifacts, training, and usage-cost questions.
Students should ask the instructor whether and how AI may be used for a particular course or assignment.
Contact the appropriate Mercer office for privacy, FERPA, HIPAA, research compliance, records, legal, accessibility, copyright, or other policy questions.
12. Use Amplify Expert for platform help
If you need help using a feature in AmplifyAI, open the Assistant Gallery and select Amplify Expert. This assistant is designed to help users with virtually any feature or function in AmplifyAI.
Use it when you want step-by-step guidance, explanations of tools or settings, or help finding the best way to complete a task in the platform.
