Transitioning from Amplify Gen AI to Google Gemini at Davidson College

Amplify Gen AI is being decommissioned at Davidson College, with a final decommissioning date of July 15th. Google Gemini is the recommended replacement for everyday generative AI work. This guide walks you through what changes, what stays the same, and how to bring your important Amplify work with you.

Table of Contents

1. At a Glance
2. Where You Go and How You Sign In
3. Feature-by-Feature: Amplify → Gemini
4. Bringing Your Important Chats With You
5. Rebuilding Your Amplify Assistants and Prompt Templates in Gemini
6. What Gemini Brings
7. Good to Know
8. Tips for Your First Week in Gemini
9. Getting Help

1. At a Glance

Gemini centers on a single family of models with two everyday modes — Fast for quick work and Thinking for deeper reasoning — plus Gems, a powerful way to save instructions, files, and context into a reusable AI helper you can return to (and share) anytime.

2. Where You Go and How You Sign In

Amplify (legacy): amplify.davidson.edu → sign in with your Davidson email and password.

Gemini (replacement): gemini.google.com → sign in with your Davidson Google account (the same one you use for Gmail and Drive). You'll know you're on Davidson's instance when you see a shield icon next to the prompt box on the Gemini homepage — that shield confirms Davidson's institutional data protections are in effect, the same protections that apply to your Davidson Google Drive.

NotebookLM lives separately at notebooklm.google.com — sign in with the same Davidson Google account.

3. Feature-by-Feature: Amplify → Gemini

Choosing a Mode: Fast or Thinking

Gemini offers two everyday modes you can switch between when you start a chat:

• Gemini Fast is built for speed and efficiency. It's the right starting point for most everyday work — quick questions, brainstorming, drafting short emails, summarizing simple texts, light editing. Fast also has generous usage limits, so you can use it freely.
• Gemini Thinking is built for depth. Reach for it when you're working on something that benefits from the model taking its time: detailed research, long-form writing where coherence matters, strategic planning, complex logic, or anything where you want the AI to reason carefully through a multi-step problem.

How to decide: start in Fast. Switch to Thinking when a response feels shallow, gets the logic wrong, loses the thread in a long conversation, or glosses over the hard part of your request — give Thinking the same prompt and compare. A common pattern is to draft in Fast and refine in Thinking.

Thinking has stricter usage limits than Fast, so save it for the work that actually benefits from it.

Custom AI Helpers: Assistants → Gems

Amplify's Assistants and Gemini's Gems play the same role: a saved, named AI helper pre-loaded with instructions and (optionally) reference files. The mental model is identical — only the name and the interface differ.

To create a Gem: go to gemini.google.com, find Explore Gems or the Gem manager in the left sidebar (it may look like a jewel icon), click New Gem, name it, write its instructions, optionally attach files, test it in the preview pane on the right, and save.

One nice difference: when you attach Google Docs, Sheets, or Slides to a Gem, the files stay live-linked — edits to the source document show up in the Gem automatically.

Reusable Instructions: Prompt Templates and Custom Instructions → Gem Instructions

Amplify split this into two features: Prompt Templates (reusable prompts with {{variable}} placeholders) and Custom Instructions (overarching behavior rules). Gemini consolidates both into the Instructions field of a Gem.

If you used Amplify Prompt Templates with variables, you can get the same effect by writing a Gem whose instructions describe the task and ask you to provide the variable inputs at the start of each chat — Gemini gathers the context conversationally rather than through template placeholders.

Sharing

Amplify lets you share conversations, Assistants, and Prompt Templates with other Davidson users. Gemini's sharing is centered on Gems: you can share a Gem with specific people (as Viewer or Editor), with an expiration date if you want, or via a copyable link.

To share a Gem: open Explore Gems, click the Share icon next to the Gem, enter Davidson email addresses, choose Viewer or Editor, and send.

Note that sharing a Gem also shares whatever files are attached to it. If the Gem includes files uploaded from your device, Gemini will create a new folder in your Google Drive to store them — so the cleanest approach is to organize files in Google Drive before you build a shareable Gem.

If you used to share Amplify chat conversations directly with a colleague, the closest Gemini equivalent is to copy the relevant content into a Google Doc and share that, or to build a Gem around the topic and share the Gem.

File Uploads

Amplify accepted file uploads (most PDFs, DOCX, CSV) within a chat to give the model context. Gemini does the same. For deeper document work — analyzing several sources, building structured outlines, generating insights grounded strictly in your uploaded materials — use NotebookLM (notebooklm.google.com). NotebookLM is the Davidson-recommended tool whenever the answer needs to come from your documents rather than from the model's general knowledge.

4. Bringing Your Important Chats With You

Amplify does not have a built-in "export" feature in the support documentation. The reliable way to preserve a conversation you want to keep is to copy it out manually before Amplify is retired, then re-introduce it to Gemini as context when you need it.

Step 1 — Identify which chats are worth saving

You probably don't need to save everything. Good candidates:

• Conversations where you and the model worked out something hard (a tricky outline, a debugging session, a research synthesis).
• Conversations that contain reference material you'd want to ground future work in.
• Conversations connected to ongoing projects.

You can probably let go of one-off quick questions and exploratory brainstorms whose outputs you already used.

Step 2 — Save each important chat to a file

In Amplify, open the conversation, select the messages you want to keep, and copy the contents. Paste into one of the following:

• A Google Doc (recommended — lives in Drive, easy to attach to a Gem later).
• A plain text file (.txt) if you prefer something lightweight.
• A Word document (.docx) if you want to keep formatting and the file is going to live somewhere besides Drive.

Give the file a descriptive name (for example, Amplify Chat — Course Redesign Brainstorm — Jan 2026) and save it to a clearly-named folder in your Davidson Google Drive (something like Amplify Archive). This makes step 3 painless.

Step 3 — Re-establish the context in Gemini

When you're ready to continue the work in Gemini, you have two paths:

One-time follow-up. Start a new chat at gemini.google.com, attach the saved file using Gemini's file-upload control, and open with a prompt that frames what you want — for example: "Attached is a conversation I had previously about [topic]. Please read it for context, then help me with [next step]."

Ongoing project. Build a Gem for the project. In the Gem's instructions, describe the goal of the work and the role you want Gemini to play. In the Gem's attached files, include the saved Amplify chat plus any other reference materials. From then on, every chat you start with that Gem starts with full context — no re-attaching needed.

For document-heavy work (multiple sources, research synthesis), upload the saved chats and supporting documents into a NotebookLM notebook instead. NotebookLM is purpose-built for answering questions grounded in a set of sources.

5. Rebuilding Your Amplify Assistants and Prompt Templates in Gemini

If you invested time building Amplify Assistants or Prompt Templates, you don't have to start from a blank page in Gemini. Here's the pattern:

For each Amplify Assistant you want to keep:

1. Open the Assistant in Amplify and copy its name, description, instructions, and disclaimer (if any).
2. Note which external data sources / uploaded files it relied on. If those files came from your computer, locate the originals. If they're already in Google Drive, even better.
3. In Gemini, create a New Gem. Paste the Amplify Assistant's instructions into the Gem's Instructions field. Adjust any references to "Amplify" or model-specific behavior.
4. Attach the same files (Google Docs/Sheets/Slides are recommended — they stay live-linked).
5. Test the Gem in the preview pane on the right side of the builder. Refine the instructions until the responses match what your Amplify Assistant produced.
6. Save.

For each Amplify Prompt Template you want to keep:

Gemini handles variable inputs conversationally rather than through template placeholders. Two options:

• Import the template into a Gem. Create a Gem whose instructions say something like: "When the user starts a chat, ask them for [list of variables]. Then perform [the task] using their inputs." This trades automatic placeholder substitution for a short conversational opening, but it works for almost every template use case.
• Keep the template text in a Google Doc and paste it (with your variables filled in) at the start of a chat when you need it.

6. What Gemini Brings

A few of Gemini's standout capabilities:

• Native multimodal input and output. Gemini works fluently with text, code, images, audio, and video — drop in a screenshot, a recording, or a PDF and ask Gemini to analyze it directly.
• Web search. Gemini can search the web to ground answers in current information — useful for recent events, fresh research, or anything where you want the AI's response anchored to live sources rather than just its training data.
• Google Workspace integration. Gemini works directly with content from Google Docs and Drive — pull in a doc as context, generate a draft into a doc, or attach Drive files as live-linked context to a Gem. (Direct integration with Google Sheets is not yet available.)
• NotebookLM for deep research. Upload multiple sources and ask grounded questions against just those sources — ideal for literature reviews, course prep against assigned readings, or analyzing a set of policy documents. Visit notebooklm.google.com to get started.
• A dedicated mobile app for working with Gemini away from your computer.

7. Good to Know

A few things worth being aware of as you settle in:

• Usage limits. Gemini Fast has generous limits; Gemini Thinking has stricter ones. If you hit a Thinking limit, capacity restores 24 hours later.
• Expect to iterate on prompts. Every AI tool has its own personality — treat the first response as a draft and refine from there. If a prompt isn't landing in Fast, try it in Thinking, or rephrase.
• Verify factual claims. Treat outputs as drafts to verify, especially for anything you'll publish or cite. When you do need grounded answers, Gemini's web search and NotebookLM are your best tools.
• Each chat starts fresh. New Gemini chats don't carry memory from previous ones. Use Gems with attached files when you need persistent context across sessions.

8. Tips for Your First Week in Gemini

• Start with Gemini Fast. It's quick and has generous limits — great for getting comfortable with the interface.
• Try one task you used to do in Amplify and do it in Gemini. Drafting an email, summarizing a PDF, brainstorming an outline. Notice what feels different.
• Build one Gem. Pick a task you do often (drafting weekly updates, generating alt text, writing meeting notes) and turn it into a Gem. This is where Gemini starts saving you real time.
• Try NotebookLM with one project's source documents. Even a single notebook with three or four PDFs will show you what the tool is good at.
• Be specific in your prompts. This advice applied in Amplify and still applies here. Clear, detailed prompts get better answers.

9. Getting Help

If something isn't working or you need help transitioning a specific Amplify workflow, T&I is happy to help. Contact us at ti@davidson.edu. 

Related Articles

• Getting Started with Google Gemini at Davidson College
• Amplify Gen AI vs. Google Gemini at Davidson College
• Getting Started with Amplify Gen AI at Davidson College (legacy)
• Amplify Gen AI: Advanced Use Guide for Davidson College (legacy)