Paramclasses Academy - Blog

Tutorials Details

←
Post 9 of 20
→

How to use Gemini Apps, Extensions and Custom Workflows ?

How to use Gemini Apps, Extensions and Custom Workflows ?

October 22, 2026

Module 9: Gemini Apps, Extensions and Custom Workflows

Module Overview

Google Gemini offers a range of AI features that can support everyday productivity, personalized assistance, research, and connected workflows. Depending on the features available to your account, you may be able to organize conversations, personalize responses, create custom assistants with Gems, conduct structured research, and connect Gemini with supported applications.

This module explains how these capabilities work, how to use them effectively, and what to check before relying on their outputs.

Important: Gemini features, names, availability, usage limits, and connected-app capabilities can change over time. Access may depend on your account, region, subscription, age, administrator settings, and the version of the product you use. Always check the controls and options displayed in your own Gemini interface.

Learning Objectives

By the end of this module, you will be able to:

  • Understand important features of the Google Gemini app.

  • Manage conversations and review available activity and personalization controls.

  • Create and use custom AI assistants with Gemini Gems.

  • Plan and conduct structured research using Gemini Deep Research when available.

  • Understand how connected apps and extensions work.

  • Design repeatable AI workflows for learning, content creation, research, and business tasks.

  • Apply privacy, accuracy, and security practices when using connected AI tools.


Topic 33: Google Gemini App Features Explained: Chats, Saved Activity and Personalization

1. Introduction to the Gemini App

Google Gemini is an AI assistant that can help users answer questions, generate content, summarize information, brainstorm ideas, explain concepts, and work with supported inputs.

The Gemini app provides an interface for interacting with AI models and accessing the features available to your account.

Depending on the interface and account configuration, you may encounter features such as:

  • Conversations and chat history.

  • Follow-up questions.

  • File or image inputs.

  • Saved preferences or personalization controls.

  • Activity and privacy settings.

  • Custom assistants and connected services.

The exact feature set can vary, so it is important to distinguish between capabilities that are available in general and features that are enabled for your particular account.

2. Understanding Gemini Conversations

A conversation is a sequence of messages between you and Gemini.

Conversations are useful when a task requires multiple steps. For example, you might first request a website outline, then ask for improved headings, and finally request a revised draft.

Step 1: Start a conversation

Open the Gemini app or website available to your account and begin a new conversation.

Describe the task you want to complete. Include the purpose, audience, relevant context, and expected output.

Step 2: Review the response

Check whether the answer meets your requirements. Look for missing details, unsupported claims, formatting problems, or assumptions.

Step 3: Ask a follow-up question

Use the same conversation to refine the result.

For example:

  • "Make the explanation easier for beginners."

  • "Convert the answer into a numbered checklist."

  • "Add a practical example for a small business."

  • "Identify which claims need verification."

  • "Keep the existing structure but improve clarity."

Step 4: Start a new conversation when appropriate

Use a separate conversation when you want to work on an unrelated task or avoid carrying irrelevant context into a new discussion.

3. Understanding Chat History

Depending on your account and settings, Gemini may provide a way to review previous conversations or recent activity.

Chat history can help you return to previous work, revisit useful explanations, and continue unfinished tasks.

However, chat history, saved preferences, and activity records are not necessarily the same thing. They may have separate controls and retention behavior.

Before sharing sensitive information, understand the activity and privacy settings applicable to your account.

4. Understanding Saved Activity and Privacy Controls

Activity settings can influence what interactions are saved and which related features may be available.

The exact controls and their effects depend on the current product interface and applicable account policies.

A useful privacy review includes:

  1. Open the account or settings menu.

  2. Find the available activity and privacy controls.

  3. Read what each control actually covers.

  4. Review any available saved activity.

  5. Adjust settings according to your needs.

  6. Check whether the setting applies to future activity, existing records, or both.

Do not assume that turning off one activity setting automatically deletes every existing record or affects all connected Google services.

If you use a managed work or school account, an administrator may control some settings.

5. Personalization in Gemini

Personalization can help an AI assistant tailor responses to your preferences, needs, or working style when supported and enabled.

For example, a user may prefer:

  • Beginner-friendly explanations.

  • Concise summaries.

  • Detailed step-by-step instructions.

  • Examples related to web development.

  • Tables for comparisons.

  • Professional language for business writing.

A useful personalization instruction might be:

"Explain technical concepts in clear, beginner-friendly English. Use practical examples, define unfamiliar terminology, and separate verified facts from assumptions."

Personalization should improve the relevance of responses without replacing the need to check accuracy.

Personalization best practices

  • Specify useful preferences rather than sharing unnecessary personal information.

  • Review the controls available to your account.

  • Avoid storing confidential business information as a general preference.

  • Reassess preferences if the assistant starts producing unsuitable answers.

  • Remember that personalization does not guarantee correct or unbiased responses.

6. Practical Example: Create a Repeatable Learning Workflow

Suppose you are learning JavaScript and want consistent explanations.

Use this prompt:

"Act as a JavaScript learning assistant. For every topic, provide:

  1. A simple definition.

  2. A step-by-step explanation.

  3. One beginner-friendly code example.

  4. The expected output.

  5. Two common mistakes.

  6. Three practice questions.

  7. A short revision summary.

Use clear English. Do not claim that code has been executed unless it actually has been tested."

You can reuse this prompt in future conversations or incorporate suitable instructions into a custom assistant when that feature is available.

7. Practical Exercise

Explore the Gemini interface available to you.

  • Start a conversation about a topic you are learning.

  • Ask two follow-up questions.

  • Find the available conversation-history controls.

  • Review the activity and personalization settings.

  • Record which features are available in your account.

  • Write a short note explaining which controls you would use before handling sensitive information.

Key Takeaway

Understanding conversations, activity controls, and personalization helps you use Gemini more effectively. It also helps you make informed decisions about what information you share and how your AI interactions are managed.


Topic 34: How to Use Gemini Gems to Create Custom AI Assistants

1. What Are Gemini Gems?

Gemini Gems are customizable AI assistants designed to help with specific tasks or recurring workflows.

Instead of repeatedly explaining the same requirements in every conversation, you can configure a Gem with suitable instructions and use it for related tasks.

Examples include:

  • A coding tutor.

  • A content-writing assistant.

  • A study planner.

  • A marketing brainstorming assistant.

  • A document-review assistant.

  • A research-planning assistant.

The exact configuration options, sharing features, and availability may vary by account.

A Gem is not a guarantee that every answer will be accurate. It follows its instructions within the capabilities and limitations of the underlying Gemini experience.

2. When Should You Create a Gem?

A Gem is useful when a task has a repeatable structure.

For example, if you regularly create educational articles, you may want every article to include:

  • A clear introduction.

  • Learning objectives.

  • Step-by-step explanations.

  • Practical examples.

  • Common mistakes.

  • Exercises.

  • A conclusion.

A reusable Gem can help apply this structure consistently.

For a one-time question, a normal conversation may be sufficient. You do not need a custom assistant for every task.

3. How to Create a Gem

If Gem creation is available in your Gemini interface, follow these general steps.

Step 1: Open the Gems area

Open Gemini and locate the section for Gems or custom assistants.

Step 2: Start creating a Gem

Select the available option to create a new Gem.

Step 3: Choose a clear name

Use a name that describes the assistant's purpose.

Examples:

  • JavaScript Learning Coach.

  • SEO Content Planner.

  • Business Email Assistant.

  • Research Outline Builder.

Step 4: Write specific instructions

Explain the assistant's role, task, audience, output structure, and restrictions.

Avoid vague instructions such as "Be an expert at everything."

Step 5: Add supporting knowledge if supported

If the interface offers a way to attach reference files, provide relevant and appropriate documents.

Use materials that you have permission to share. Avoid confidential or unnecessary personal information.

Step 6: Save and test the Gem

Use a realistic example task to check whether the assistant follows the instructions.

If the output is inconsistent, improve the instructions and test again.

4. The Five-Part Gem Instruction Formula

A practical Gem instruction can be organized into five parts.

Part 1: Role

Define what the assistant should do.

Example: "You are a beginner-friendly JavaScript tutor."

Part 2: Audience

Specify who will use the output.

Example: "The learner has basic computer knowledge but little programming experience."

Part 3: Workflow

Define the sequence of work.

Example: "Explain the concept, show code, explain the output, and then provide practice questions."

Part 4: Output format

Specify the required structure.

Example: "Use headings, code blocks, numbered steps, and a short recap."

Part 5: Quality and safety rules

Explain how the assistant should handle uncertainty and mistakes.

Example: "Do not invent test results. Explain assumptions and recommend testing code before use."

5. Practical Example: Create a JavaScript Tutor Gem

Use the following instructions when configuring your Gem.

Gem Name: JavaScript Learning Coach

Gem Instructions:

"You are a patient JavaScript tutor for beginners.

Your purpose is to help learners understand JavaScript concepts through simple explanations and practical exercises.

For every topic:

  1. Begin with a short definition.

  2. Explain the concept step by step.

  3. Provide a small, readable JavaScript example.

  4. Explain the important lines of code.

  5. Show the expected output.

  6. Identify common beginner mistakes.

  7. Provide three practice questions.

  8. Include answers in a separate answer section when requested.

Use clear English and define unfamiliar technical terms.

Do not assume the learner has advanced programming knowledge. Avoid unnecessary jargon.

If a code example has not been executed, do not claim that it has been tested. Identify assumptions and recommend appropriate tests.

If the learner provides code, explain its behavior before suggesting changes. Preserve the intended functionality unless a change is necessary.

Prioritize correctness, clarity, accessibility, and safe programming practices."

Test the Gem

After saving it, ask:

"Teach me JavaScript variables using let, const, and simple examples."

Check whether the response includes the required explanation, examples, common mistakes, and exercises.

If the Gem misses important requirements, revise its instructions rather than assuming the first configuration is perfect.

6. Practical Example: Create a Marketing Assistant Gem

Gem Name: Digital Marketing Content Planner

Gem Instructions:

"You are a digital marketing planning assistant for small businesses.

Help create content ideas, campaign outlines, social media calendars, and draft marketing copy.

Before producing a campaign, identify the target audience, campaign objective, platform, desired tone, and call to action. If essential information is missing, ask a focused clarification question.

When creating a content calendar, include the date or proposed publishing day, topic, content format, key message, and suggested call to action.

When writing marketing copy, avoid unsupported claims, fabricated testimonials, guaranteed results, and misleading statistics.

Distinguish creative suggestions from verified facts. Do not claim that a campaign will guarantee sales, ranking improvements, or engagement.

Use a professional, practical, easy-to-edit format."

Example request

"Create a seven-day social media content plan for a fictional web development training institute. Focus on beginner-friendly learning, practical projects, and student engagement. Include the topic, format, key message, and CTA for each day."

Review the resulting calendar and revise it to fit your actual audience and publishing schedule.

7. How to Improve a Gem

Use the following process:

  1. Test the Gem with a realistic task.

  2. Compare its output against your requirements.

  3. Identify which instruction was missing or unclear.

  4. Update the instructions.

  5. Test the same task again.

  6. Test a different task to check whether the improvement generalizes.

Do not add unnecessary rules that make the Gem difficult to use. Focus on instructions that consistently improve output quality.

8. Limitations and Best Practices

  • A Gem may misunderstand instructions.

  • It may produce inaccurate information.

  • It may not have access to every file or connected service.

  • Its behavior depends on the model and features available to your account.

  • A Gem should not replace human review for high-impact decisions.

  • Uploaded reference files may contain sensitive information and should be selected carefully.

  • Sharing a Gem or its supporting material may expose information depending on the sharing configuration.

Practical Exercise

Create one Gem for a recurring task.

Choose either:

  • A study assistant.

  • A coding tutor.

  • A writing assistant.

  • A research planner.

  • A marketing assistant.

Give it a clear role, audience, workflow, output format, and quality rules.

Test it with three different prompts and record which instructions need improvement.

Key Takeaway

Gems can make repeated AI tasks more consistent by packaging instructions into a reusable assistant. Their usefulness depends on the quality of the instructions, appropriate access to supporting information, and regular review of the outputs.


Topic 35: How to Use Gemini Deep Research for Structured Research Tasks

1. What Is Gemini Deep Research?

Gemini Deep Research is a research-oriented capability designed to help with more involved information-gathering tasks, where supported by the user's account and current product interface.

Instead of answering only a short question, a research workflow may organize a topic into subquestions, gather information from available sources, and produce a structured report.

It can be useful for:

  • Exploring an unfamiliar industry.

  • Comparing technologies.

  • Preparing a research brief.

  • Investigating a complex topic.

  • Collecting background information for an article.

  • Developing an initial literature or source review.

The exact workflow, available sources, and output capabilities may change over time.

Important: A polished research report is not automatically accurate or comprehensive. Source quality, date, relevance, and claim verification remain essential.

2. When Should You Use Deep Research?

Use a research-oriented workflow when the task requires multiple sources, several subquestions, or a structured synthesis.

Examples:

  • Comparing different website development frameworks.

  • Studying how AI tools are used in education.

  • Reviewing current approaches to cybersecurity awareness.

  • Exploring the competitive landscape of a business sector.

For a simple definition or straightforward question, a normal Gemini conversation may be faster and sufficient.

3. Define the Research Question

Before starting, identify:

  • Main question: What do you need to learn?

  • Scope: Which topic, region, industry, or period matters?

  • Audience: Who will use the report?

  • Evidence requirements: Which types of sources are appropriate?

  • Deliverable: Do you need a summary, comparison, report, or recommendations?

A broad question such as "Research AI" is difficult to evaluate.

A more useful question is:

"Prepare a structured overview of how generative AI is used in beginner programming education. Cover common applications, potential benefits, limitations, privacy concerns, and practical evaluation criteria. Prioritize credible educational research and official documentation where relevant."

4. How to Start a Deep Research Task

If Deep Research is available in your Gemini interface, use the following general workflow.

Step 1: Open the available research feature

Locate the Deep Research option or the equivalent research workflow shown in your interface.

Step 2: Enter a specific research prompt

Define the research question, scope, audience, and required report structure.

Step 3: Review the research plan if provided

Some research workflows may present a plan or outline before conducting the full task.

Check whether the proposed subquestions cover your actual objective. Refine the scope if the plan is too broad or omits an important issue.

Step 4: Review the resulting report

Examine the main findings, supporting sources, dates, assumptions, and any limitations stated in the report.

Step 5: Verify important claims

Open the cited sources when possible. Confirm that they actually support the claims made in the report.

Step 6: Produce the final deliverable

Use the research as a starting point for a report, presentation, article outline, or decision brief. Clearly distinguish verified findings from recommendations and unresolved questions.

5. Practical Example: Research Website Development Frameworks

Use this prompt:

"Research the current landscape of widely used front-end web development frameworks and libraries for building modern websites.

Scope:

  1. Compare React, Vue, Angular, and Svelte.

  2. Explain their general approaches and common use cases.

  3. Compare learning considerations, ecosystem maturity, development workflow, and maintenance considerations.

  4. Identify situations where plain HTML, CSS, and JavaScript may be sufficient.

  5. Distinguish official documentation from third-party opinions.

  6. Use current, relevant sources and include publication or update dates where available.

  7. Cite sources for factual claims and identify any claims that could not be independently verified.

  8. Present the findings in a comparison table followed by a balanced conclusion.

Do not declare one technology universally best. Make recommendations conditional on project requirements, team experience, and maintenance needs."

What should you evaluate?

Check whether the report:

  • Uses relevant and recent sources.

  • Distinguishes frameworks from libraries accurately.

  • Provides evidence for its comparisons.

  • Avoids unsupported popularity or performance claims.

  • Recognizes that project requirements affect the best choice.

  • Separates facts from subjective recommendations.

6. Evaluate Sources Critically

Not all sources are equally reliable.

A practical source-evaluation checklist includes:

Authority: Who published the information, and what expertise do they have?

Evidence: Does the source provide data, documentation, methodology, or verifiable references?

Recency: Is the information current enough for the question?

Relevance: Does the source directly address the research question?

Bias: Could commercial, political, or personal interests influence the presentation?

Traceability: Can you locate the original source and verify the relevant claim?

Official documentation is often appropriate for product behavior and technical specifications. Independent research and credible reporting may be better suited to questions about effectiveness, market trends, or public impact.

7. Use a Claim-Verification Table

For important research projects, create a table that links claims to evidence.

Research claim Supporting source Verification status
Claim A Original source or official documentation Verified / Needs review
Claim B Relevant study or report Verified / Needs review
Claim C No adequate source found Unverified

Replace the example labels with actual claims and sources from your research.

Do not mark a claim as verified merely because Gemini included a citation. Open the source and check the supporting material.

8. Ask Gemini to Identify Research Gaps

After receiving a report, use this follow-up prompt:

"Review the report and identify important questions that remain unanswered. List unsupported claims, outdated sources, conflicting findings, and missing perspectives. For each issue, explain what additional evidence would be needed. Do not invent sources or present unresolved questions as established facts."

This can help reveal where further research is required.

9. Turn Research into a Practical Deliverable

Once the research has been checked, ask Gemini to transform it into a suitable format.

For example:

"Using only the findings supported by the reviewed sources, create a concise decision brief with an executive summary, key findings, comparison table, risks, recommendations, and open questions. Preserve source references next to relevant claims. Clearly label recommendations as recommendations rather than established facts."

Always verify that the final document retains the source context and does not introduce unsupported claims during summarization.

10. Limitations and Best Practices

  • Deep Research may not find every relevant source.

  • Sources can be outdated, biased, or incorrectly interpreted.

  • Citations may not fully support the accompanying claim.

  • Research reports can omit important counterarguments.

  • Availability and usage limits may differ by account.

  • Sensitive or confidential research material should be handled carefully.

  • High-impact legal, financial, medical, or security decisions require appropriate expert review.

Practical Exercise

Choose a topic related to education, web development, business, or AI.

Prepare a research question with a clear scope. Use the available research workflow to create a structured report, then:

  1. Inspect at least five relevant source references if the workflow provides them.

  2. Verify several important claims against the original sources.

  3. Identify at least two limitations or unresolved questions.

  4. Create a one-page summary.

  5. Separate verified findings from your own recommendations.

Key Takeaway

Deep Research can help organize complex research into a structured workflow. The quality of the final result depends on the research question, the quality of sources, and careful verification of the evidence.


Topic 36: Google Gemini Extensions and Connected Apps: What They Do and How to Use Them

1. What Are Extensions and Connected Apps?

Connected apps allow an AI assistant to work with supported external services or information sources, subject to the access and permissions available in a particular product configuration.

Depending on the current Gemini experience, these capabilities may be presented through connected apps, extensions, integrations, or similar interface options.

Potential use cases include:

  • Finding relevant information in supported services.

  • Working with information from connected productivity tools.

  • Supporting travel or planning tasks.

  • Retrieving relevant information from supported personal content.

  • Combining AI assistance with existing digital workflows.

The exact services and available actions vary by account, region, administrator settings, and product version.

Do not assume that every Google service or third-party application is available in every Gemini experience.

2. Why Use Connected Apps?

Without a connected service, you may need to copy information manually into a prompt.

When an appropriate connection is available and authorized, a workflow may allow Gemini to use information from that service without requiring you to paste every relevant detail.

However, access is not necessarily unlimited. A connection may support only certain information types or actions.

For example, permission to search or summarize information does not automatically mean that the assistant can send messages, modify files, or perform every action within the connected service.

3. How to Check Available Connections

Use this general process, adapting it to the options shown in your interface.

Step 1: Open Gemini settings or connected-app controls

Look for a section that lists connected apps, extensions, integrations, or available services.

Step 2: Review available services

Check which services are listed and whether they are enabled for your account.

Step 3: Read the permissions

Understand what information or actions a connection may access.

Step 4: Connect or enable a service if appropriate

Follow the consent and authorization steps displayed by the product.

Do not grant access to information that is unnecessary for your task.

Step 5: Test a low-risk request

Start with a simple task, such as locating a non-sensitive document or summarizing a sample of information that you are comfortable sharing.

Step 6: Verify the result

Check whether the response reflects the correct source, account, and information.

If a feature is unavailable, do not assume it is a technical fault. It may not be supported for your account or current configuration.

4. Practical Example: A Connected-Information Workflow

Suppose you need to prepare for a meeting and have relevant notes in a supported connected service.

If your Gemini account supports access to that service, try a request such as:

"Using the relevant meeting notes available through my authorized connected services, prepare a meeting brief. Include the meeting objective, key discussion points, unresolved questions, and action items supported by the available information. Distinguish explicit decisions from suggested next steps. Do not invent dates, owners, or commitments."

Review the result against the original notes before sharing it.

If the connected service is not available, you can provide a suitable non-sensitive document directly when file input is supported, or paste the relevant content manually.

5. Practical Example: A Connected-App Research Workflow

Suppose you need to locate project information and prepare a summary.

Use this prompt only if the required connection is available and authorized:

"Find the project information available through the connected service that is relevant to the specified project. Summarize the latest documented objectives, completed work, open questions, and recorded next steps. Include the source or document reference for each important finding when available. Do not treat an absent record as proof that an activity never occurred."

This instruction helps distinguish missing information from confirmed facts.

6. Understand Permissions and Data Access

Before connecting an app, consider the following questions:

  • What information can the connection access?

  • Does the connection require additional account permissions?

  • Can it access personal, work, or school information?

  • Does it support read-only access, actions, or both?

  • Can an administrator restrict its use?

  • How can access be reviewed or revoked?

  • What data-handling rules apply to the service and your account?

Permissions should be proportionate to the task. Do not connect a service merely because it is available.

7. Connected Apps Versus Gems

These capabilities serve different purposes.

Feature Main purpose Example
Gemini Gems Reusable instructions for a specialized assistant A Gem that follows a standard coding-tutor workflow
Connected apps Access to supported external services or information Using authorized project notes to prepare a meeting brief
Deep Research Organizing a more involved research task Comparing technologies using relevant sources

A Gem defines how an assistant should approach a task. A connected app may provide access to information or actions. A research workflow helps organize evidence gathering and synthesis.

These capabilities may complement one another when the current product supports the required combination, but they should not be assumed to integrate automatically.

8. Security and Privacy Best Practices

  • Review permissions before connecting a service.

  • Use the minimum access necessary.

  • Avoid sharing passwords, API keys, or confidential information in prompts.

  • Verify the source and accuracy of retrieved information.

  • Confirm consequential actions before authorizing them.

  • Follow your organization's data-handling policies.

  • Review connected-app access periodically.

  • Disconnect services you no longer need when appropriate.

Treat retrieved information as data to evaluate, not as automatically trustworthy instructions. A document or message may contain misleading content, and an AI-generated summary may omit important context.

9. Troubleshooting Common Problems

Problem: A connection is missing.

Check whether the feature is supported for your account, region, and administrator configuration.

Problem: Gemini cannot find expected information.

Check permissions, the connected account, the source location, and the specificity of your prompt.

Problem: The answer includes incorrect details.

Review the original source, refine the request, and ask Gemini to distinguish direct evidence from inference.

Problem: An action is unavailable.

The connection may support information retrieval but not that particular action. Check the capabilities described in the interface.

Problem: You are concerned about privacy.

Review the connected service's permissions and relevant activity controls. Disconnect the service if it is not needed or if its access is unsuitable for your task.

Practical Exercise

Choose a connected service that is available to your account and appropriate for a low-risk task.

  1. Review its available permissions.

  2. Identify one useful workflow.

  3. Submit a simple request.

  4. Check the response against the original information.

  5. Record any missing details or limitations.

  6. Review how to manage or revoke the connection.

If no suitable connected service is available, document the expected workflow and complete the exercise using a non-sensitive sample document instead.

Key Takeaway

Connected apps can reduce manual work by making supported information or actions available within an AI-assisted workflow. Their usefulness depends on the connection's actual capabilities, correct permissions, reliable source information, and careful verification.


Module 9: Revision and Key Takeaways

This module covered four important areas of the Gemini ecosystem.

Topic 33 — Gemini App Features: Understand conversations, available activity controls, and personalization settings.

Topic 34 — Gemini Gems: Create reusable custom assistants with clear roles, workflows, output formats, and quality rules.

Topic 35 — Deep Research: Structure complex research tasks, evaluate source quality, verify claims, and turn findings into practical reports.

Topic 36 — Extensions and Connected Apps: Understand authorized connections, permissions, information access, and the limitations of integrated workflows.

Recommended Workflow for Using These Features Together

  1. Define the task and the expected outcome.

  2. Use a Gem if the task follows a recurring pattern.

  3. Use a connected app only when relevant information is available and access is appropriate.

  4. Use a research workflow when multiple sources and subquestions are required.

  5. Verify important facts against original sources.

  6. Review privacy, permissions, and the final deliverable.

Not every task needs every feature. Choose the simplest workflow that meets the requirement.


Module 9 Assessment

Choose the best answer for each question.

1. What is the primary purpose of a Gemini Gem?

A. To guarantee that every AI response is correct
B. To provide a reusable assistant configured for specific tasks
C. To replace every connected application
D. To automatically verify every online source

2. Why should users review activity and privacy settings?

A. Every setting affects all Google services in exactly the same way
B. Reviewing settings helps users understand available controls and how their interactions may be handled
C. Turning off one setting always deletes all historical data
D. Privacy settings are relevant only to developers

3. What makes a good Gem instruction?

A. A vague request to be an expert at everything
B. A role, audience, workflow, output format, and quality rules
C. A list of unrelated keywords
D. A requirement to always agree with the user

4. When is a research-oriented workflow most useful?

A. When a complex topic requires multiple sources and structured synthesis
B. When a user needs only a simple definition
C. When source verification is unnecessary
D. When a report must contain invented citations

5. Why should you open and verify research sources?

A. A citation automatically proves every claim
B. A polished report cannot contain errors
C. Sources may not fully support the claims attributed to them
D. Research reports never include outdated information

6. What is a key consideration when connecting an app?

A. Granting every permission by default
B. Reviewing the service's access and permissions
C. Assuming all connected apps support identical actions
D. Ignoring the account being connected

7. Which statement about Gems and connected apps is correct?

A. Gems and connected apps are exactly the same feature
B. Gems provide reusable instructions, while connected apps may provide access to supported services or information
C. Connected apps guarantee factual accuracy
D. Gems automatically gain access to every file on a device

8. What should you do if a connected-app response contains incorrect information?

A. Accept the response because it came from an integration
B. Assume the original source is always wrong
C. Check the original source, permissions, and prompt
D. Remove all source references

Answer Key

  1. B — A reusable assistant configured for specific tasks.

  2. B — Settings help users understand available controls and data handling.

  3. B — A clear role, audience, workflow, output format, and quality rules.

  4. A — Multi-source tasks benefit from structured research.

  5. C — Citations must be checked against the claims they support.

  6. B — Review access and permissions before connecting a service.

  7. B — Gems define reusable instructions; connected apps may provide authorized access.

  8. C — Verify the source and refine the request as needed.


Final Practical Project: Build a Personalized Gemini Workflow

Project Objective

Design a reusable AI workflow for one realistic task: learning, content creation, business research, or project coordination.

The workflow should use only the features actually available to your account and should preserve privacy and verification requirements.

Project Requirements

Part 1: Define the task

Write a clear task statement that explains:

  • The goal.

  • The intended audience.

  • The information required.

  • The expected output.

  • The criteria for a successful result.

Part 2: Configure a reusable assistant

If Gems are available, create one with:

  • A clear role.

  • Audience-specific instructions.

  • A repeatable workflow.

  • A defined output structure.

  • Rules for accuracy, uncertainty, and privacy.

If Gems are unavailable, save the instructions in a reusable prompt document.

Part 3: Identify relevant information sources

Determine whether the task needs a connected app, supplied documents, or public research.

Use only appropriate sources and permissions. Do not connect unnecessary services.

Part 4: Conduct the task

Run the workflow using a realistic example. Review the output and identify missing details, unsupported assumptions, or errors.

Part 5: Verify the results

Check important claims against their sources. If calculations, dates, or recorded decisions are involved, confirm them independently where appropriate.

Part 6: Document the workflow

Prepare a short report containing:

  • The task and objective.

  • The instructions used.

  • The features used or considered.

  • The final output.

  • Errors or limitations discovered.

  • Privacy and permission considerations.

  • Improvements for the next run.

Completion Checklist

  • The workflow has a clearly defined objective.

  • Reusable instructions are specific and testable.

  • Any connected service is appropriate and authorized.

  • Important factual claims are verified.

  • Personal and confidential information is handled carefully.

  • Limitations are documented.

  • The final output meets the original requirements.

  • The workflow can be repeated without unnecessary steps.

Conclusion

Google Gemini's app features, Gems, Deep Research, and connected apps can support a wide range of recurring workflows. The most effective approach is to choose the right feature for the task, provide clear instructions, control access to information, and verify important results.

By completing this module, you have learned how to build reusable AI workflows while maintaining control over accuracy, permissions, privacy, and output quality.

←
Post 9 of 20
→