Congratulations — you’ve made it to Day 7 of the AI Task Helper Agent (Gemini Edition) series!
By now, you’ve built:

• A Gemini-powered planner
• A reasoning system with multi-step logic
• A command-line interface
• A simple but effective memory system

Today, you’ll take the final step:
Turning this project into your own custom agentic AI.

This is where your creativity takes over.
Whether you want to build a Fitness Planner, Study Coach, Content Calendar Agent, Meal Prep Assistant, or Productivity Coach — today’s tutorial gives you a template to build ANY agent.

Let’s turn your codebase into a launchpad.

---

Goal

Use the existing codebase as a template to build your own agent:

• Fitness planning agent
• Study planner
• Content creation assistant
• Meal prep planner
• Productivity or habit coach
• Anything you want

Today is about customization and ownership.

---

1. Recap What You Have Built So Far

Before customizing, let’s appreciate everything your agent can already do:

Gemini-powered planner

Your agent communicates with Google’s Gemini model using smart prompts.

Advanced reasoning

You already built a two-step planning pipeline (phases + expansion).

CLI tool

You can run your agent like a real application:

python -m src.main "Plan my workout week"

Memory

Your agent stores past plans and can show task history.

This is the perfect foundation for building any specialized agent.

---

2. Introduce a Simple Customization Template

To build your own personalized agent:

• Change the agent’s persona
• Adjust prompt structure
• Add domain-specific rules (fitness, coding, diet, productivity, etc.)

You can keep everything inside agent.py or create a new file — whichever feels cleaner.

---

3. Create a New File or Branch (Optional)

If you want to keep their project organized, you can create:

src/fitness_agent.py

or

src/study_agent.py

Or create a new branch:

git checkout -b fitness-agent

This keeps the main repo clean while encouraging experimentation.

---

4. Customizable Template Prompt

Here is the DomainTaskAgent class you provided in your outline ().
Add it to agent.py, or put it in its own file:

from textwrap import dedent
from llm_client import generate_text

class DomainTaskAgent:
    """
    Generic template for domain-specific agents.
    You can subclass this or just copy the idea.
    """

    def __init__(self, domain_instructions: str):
        self.domain_instructions = domain_instructions

    def plan_task(self, user_goal: str) -> str:
        prompt = dedent(f"""
        You are an AI assistant that helps with this domain:

        {self.domain_instructions}

        The user will give you a goal. Your job:
        1. Clarify the goal in one sentence.
        2. Break it into a realistic, safe, step-by-step plan.
        3. Follow best practices for this domain.
        4. Explain anything that might be confusing.

        User goal:
        "{user_goal}"

        Respond with:

        Goal:
        - <one sentence>

        Plan:
        1. ...
        2. ...
        3. ...
        """).strip()

        return generate_text(prompt)

This class is your template.

You only need to edit:

• The domain_instructions
• The goal they want to plan

That’s it — instant custom agent.

---

5. Example: Fitness Agent

Here’s a complete example

Step 1 — Write domain instructions

fitness_instructions = """
You are a helpful fitness planning assistant.
Always prioritize safety and beginners' limits.
Avoid giving medical advice; suggest seeing a professional
if the user has injuries or health conditions.
"""

Step 2 — Create your fitness agent

fitness_agent = DomainTaskAgent(domain_instructions=fitness_instructions)

Step 3 — Run a sample prompt

plan = fitness_agent.plan_task("Get ready to run a 5K in 8 weeks")
print(plan)

This produces a specialized fitness training plan — not just a generic plan.

Now you can build:

• Meal prep agent
• Coding study agent
• Productivity coach
• Running plan agent
• Content creation scheduler

The possibilities are endless.

---

6. Challenge Your Readers

This is a great moment to become creators.

Pick a niche

Examples:

• Content creator agent
• Exam revision agent
• Weight-loss assistant
• Coding roadmap agent
• Daily habit coach
• Side-project builder

Write your own domain instructions

Teach the agent how to behave and what to prioritize.

Adjust prompts, responses, memory rules

Maybe their niche needs different steps or safety notes.

Publish your project

share your creation or link back your project here for everyone.

---

Helpful Reads

Beginner-friendly AI overview
https://tooltechsavvy.com/top-5-free-ai-tools-you-can-start-using-today-no-tech-skills-needed/

Boosting everyday productivity using AI
https://tooltechsavvy.com/chatgpt-for-beginners-7-easy-ways-to-boost-productivity-with-ai/

---

You Did It — You Built an Agentic AI!

During this 7-day journey, you have created:

• A functioning AI planning agent
• A reasoning-based agent
• A CLI tool
• A memory system
• A customizable agent template
• Your own specialized agent

You now understand the fundamentals of agentic AI, how to structure prompts, how to integrate Gemini, and how to build real-world AI tools.

And today, you finally gained the ability to create your own agent from scratch.