How to Build a Personalised Chatbot For Free with Google Gemini and No Coding Skills

6 min readJust now

–

Press enter or click to view image in full size

Two large colourful gems surrounded by tiny gems on a black velvet cloth (generated with Gemini); Source: Image by author

In August 2024, Google released an interesting feature on Gemini — Google’s state-of-the-art Multimodal Large Language Model (MLLM) — to allow users customize models on the platform. It’s called Gems (as you would guess, shortened from “Gemini”).

Gems was released to free subscribers in March 2025, making this tutorial useful to all parties.

In this tutorial, we will walk through how to make simple gems for your different needs and how you can make them more powerful with Google Docs.

N/B: This tutorial assumes you already have a functional Google account and already use Gemini.

Creating a Gem

Gems

A Gem is basically a Gemini model with a custom instruction to guide its interaction with a user. These instructions are defined before the interactions begin with the model. We call them System Prompts.

System prompts allow language models to assume personalities relevant to the user’s needs. This alleviates the need to build chat history with a model in order to build some level of task proficiency.

These instructions are declared once and are persistent across conversations.

Your First Gem

STEP 1: To create your first gem, visit Gemini, open the left panel with the three parallel strokes and select “Explore Gems”.

Source: Image by author

This will navigate to your Gem manager, where you will see premade Gems by Google and anyone you created, if you have already.

Press enter or click to view image in full size

Source: Image by author

STEP 2: Click the “New Gem” button. This will transfer you to the lab area, where you can customize Gems.

Press enter or click to view image in full size

Gem lab; Source: Image by author

STEP 3: Define the name, description, and instructions of your Gem.

Instructions are what Gems run on. Here is where Prompt Engineering shines. Instructions can range from short tasks to a list of multimodal tasks. Great instructions are specific and broken down.

STEP 4: Test your Gem in the preview window on the right: interact with your Gem and have a quick view of how it will respond when published.

STEP 5: Save your Gem by clicking the “save” button above the preview window.

Now you can return to the Gemini homepage and find your Gem listed.

Gems list from left pane; Source: Image by author

Docs Powered Gems

Knowledge

To extend the power of your Gems, the lab has a feature called “Knowledge”. Knowledge is where you can provide you Gems with context to improve its capacity. This could be images, documents, code, or slides.

You can even describe workflows in documents for your Gem to follow, so long as the processes fall within Gemini’s inert capabilities.

Say you want to generate photos of a certain kind, instead of always uploading the samples at every conversation, you can supply them in the knowledge is knowledge area and instruct the Gem to follow the examples in the knowledge.

Press enter or click to view image in full size

Gem lab area; Source: Image by author

Career Planner Gem

Docs is my go-to because of the flexibility of editing and its multimedia freedom. They can serve as a sort of skill for your Gems.

I made a Career Planner using Docs. Here’s how:

1. Create a new doc in Drive.
2. Describe the name and function of the Doc: This is the defined at the top of the document. It increases the Gem’s confidence that it is using the right knowledge. It also serves as a headline for you the builder to know what doc you are dealing with when you revisit in future.
3. Define the purpose of the Doc: This serves as a sort of extension of the Gem’s instructions. Here, I list what is expected of the Gem. If you expect the Gem to follow a specific order, ensure to define it here.
4. Describe output data: The career planner is essentially a workflow: a series of steps and processes to arrive at a clear path. Each step has its expected input/output, which is either a feed for, or from, the next or previous step/process in the workflow. Here, I define the expected data and output format for each step in the workflow.
5. Define the workflow: The workflow in this doc is called Questions — a list of grouped questions for the Gem to ask the user in order to understand the users desired direction.
6. Write in Markdown: To ensure the Gem understands the flow and various levels in each step, I wrote the entire process in markdown format.
7. Add Doc as Knowledge: Return to the lab area of your Gem and select “Add files” button (+) under Knowledge and select “Add from Drive”. Select your recently created doc from the pop up window. (You can define and add as many docs as your use case requires.)
8. Test: Interract with your Gem in the preview window and ensure it meets your expectations. You may edit instructions or your doc until your Gem performs optimally.
9. Save: When you are satisfied with your Gem’s performance, you can save it and begin mapping your career journey.

Markdown is a lightweight markup language with a simple syntax that uses plain text formatting for creating structured documents. It essentially employs the use of hashtags (#) to specify headings. The more the hashtags (##) the lower the level of the heading.

Google Doc with Custom Workflow; Source: Image by author

Press enter or click to view image in full size

Press enter or click to view image in full size

Press enter or click to view image in full size

Chatting with the Career Planner Gem; Source: Image by author

This prompting strategy was adapted from Anthropic’s **Skills **functionality in Claude AI (see here and here).

To have the full benefit of any knowledge, call it in the instructions window and specify how and when the knowledge should be used. This ensures that your Gem always refers to that knowledge when performing the specified task. Failing to do this may confuse the Gem, especially when supplying multiple knowledges.

Source: Image by author

You can experiment on how the Gem calls up knowledge to ensure instructional consistency, how much knowledge can be loaded per time, and the Gem’s latency due to referencing its knowledge. This is important to user experience.