FIRST OF ALL - INTRODUCTION

For a technology widely acclaimed as an assistant or the perfect productivity hack, the use of AI can sometimes feel very strenuous and ironically inefficient.

“Garbage In, Garbage Out,” they say. So, you learn as much prompt engineering as possible. When the time comes to use the learned principles, you try your best. You spend minutes or even hours typing out long, detailed instructions like, "Act as a Gen-Z Nigerian social media manager..." or "Please summarize this report in bullet points, keeping the tone professional and accessible to an undergraduate."

You hit “Send” and your AI gives you responses that make you hiss and ask: “Wetin be this?” It got something wrong, or missed an important detail, or you omitted something crucial in the prompt. Or in the worst scenario possible, it hallucinates. So you wrestle with it in frustration. You reiterate, craft your prompt again, or on rare occasions, settle for the flawed output.

Three days later, you have to repeat the process, so you do it again. That’s the digital equivalent of digging up a well and filling it back up. E no too make sense!

What if I told you that there is a gem tucked away inside Google Gemini that lets you hire a specialist—a dedicated expert trained by you—to handle all that repetitive work, permanently, and still deliver maximum quality?

This way, you don't have to remember any perfect 10-line prompt. You simply have to give your AI a permanent yet editable identity. You create the prompt ONCE, and you’re good to go forever. I’m talking about Gems.

What Are Gems?

In simple terms, a Gem is a customized version of the Gemini model that remembers a specific, detailed set of instructions forever.

An analogy that best differentiates standard AI prompting from using a Gem is different employment styles. Think of standard AI prompting like hiring a freelancer for a single gig. You explain the project, they deliver, and the next week, you have to explain the project again to someone new. Double wahala!

A Gem, on the other hand, is like hiring a full-time, salaried employee with a crystal-clear job description. You define their role—their system prompt—once, and they keep that role as their foundational identity for every single task you give them.

This initial system prompt sets the rules:

1. The Persona (Who It Is): “You are a witty, authoritative, and slightly sarcastic Nigerian food critic based in Lagos…”

2. The Task/Goal (What It Should Do): “Your primary task is to take a set of raw, unpolished tasting notes (provided by the user) and transform them into a compelling, polished, and shareable restaurant review.”

3. The Format (How to Deliver): “Always deliver the output in two distinct sections…”

4. The Limitations (What It Should Not Do): An example is: “DO NOT use generic Western-centric food terms. Instead, use evocative local expressions…”

5. Its Knowledge Base: "Reference the attached Brand Style Guide PDF for all tone and color palette decisions." This is important, but optional. For example, if you’re a content writer, you can attach your previous writing projects so that your Gem can understand your writing style better or gain more context.

Creating a Gem like this essentially sets up a persistent, long-context-memory AI agent. It saves your time, reduces errors, and dramatically improves output consistency.

KINI ANFAANI? WHAT DOES IT DO?

The application of Gems for AI users choke! It can do so many things, especially repetitive and time-consuming tasks like:

1. Content Creation: Gems guarantee brand consistency and format adherence across all channels. They permanently embed your specific voice, tone, and required formatting (e.g., use of local slang, hashtag rules, image description style) into every draft.

2. Business Tasks: For businesses, Gems can help you enforce structural routine. You define the exact framework (e.g., PESTLE, SWOT, monthly reporting template), and the Gem ensures all reports, summaries, and meeting minutes strictly follow that template every time.

3. Personal and Study: Gems can serve as a specialized tutor or planner. That thing wey you no understand because your lecturer dey use big big English? Gems can help you break it down welewele, based on the same analogies and learning steps you defined.

4. Technical Tasks: As a developer, you can use Gems to screen your codes and solve technical issues. It can eliminate variability in technical output by permanently remembering style guides, documentation requirements, and preferred coding languages.

Ati bee bee lo! (And so on, and so forth!)

How It Works (+ Free Example)

A Gem is the perfect example of Small body, big engine. It has a radically simple workflow and a surprisingly massive impact.

First, you have a choice: you can start by picking a ready-to-use expert from Google's gallery of Premade Gems (like a Career Guide or a Trip Planner) and customize it, or you can build a Gem from scratch.

If you choose to build your own, the process is straightforward:

1. You Define the Prompt: You go into the Gems builder and write the comprehensive system prompt detailing the agent's identity, rules, tone, and formatting.

2. The Gem is Born: You save this persona and give it a name. We’ll name this one "Agba Food Critique (Senior Food Critique)."

   

   Oluwakorede X Google Gems (Source: Screenshot by Oluwakorede)

3. Repetition is Eliminated: From that moment on, whenever you open this "Agba Food Critique," you don't need to re-type the role. You simply paste your data, or tell it your query, like "This is this week’s food review."

4. Long-Term Memory: Because the system prompt is always active, the Gem retains context and behavior across weeks, eliminating the frustrating need to remind the AI who it is and how it should speak. It’s the difference between coaching a beginner and delegating to a pro.

EXAMPLE: AGBA FOOD CRITIQUE

QUERY

OUTCOME

WHO YOU EPP? WHO’S IT FOR?

Gems are designed for anyone who performs repetitive, defined tasks.

• The Busy Professional: You can create a "Weekly Summary Gem" to instantly structure your meeting notes into a cohesive summary for your team every Monday morning.

• The Marketing Manager: Establish a "SEO Blog Writer Gem" to ensure all blog drafts consistently meet specific keyword density, formatting, and word count requirements.

• The Content Creator : Establish a "Brand Voice Adaptor Gem" to ensure all outputs adhere to the same brand standards and key messaging.

• The Student : Build a "Concept Simplifier Gem" that automatically processes dense reading materials or lecture transcripts and formats the output into simplified, bulleted notes or flashcards, ready for quick revision. (Note: As a student, you can maximize this tool even more with the enablement of your free Google AI Pro subscription)

• The Consultant: Design a "Client Proposal Drafter Gem" that embeds their unique value proposition and brand voice into every new proposal, regardless of the prompt.

PROs and CONs

Why You Should Consider Using Gems

1. Persistent Agent: Gems possess long-term memory for specific tasks, unlike standard chats.

2. Reduces Repetition: They eliminate the need to copy/paste or craft complex instructions daily.

3. Very Large Context: Compared to other popular models, Gems have very large context windows. This helps them to handle extensive data for specialized workflows (e.g., summarizing massive documents).

4. Specialized Workflows: You can have different Gems doing different things effectively for you. This is ideal for creating high-fidelity, highly consistent output for niche needs, as opposed to using one model for different queries in the same chat!

1. Easy Onboarding: The simple interface makes it quick for beginners to create their first specialized AI.

2. Knowledge Uploads: Can reference uploaded files (e.g., style guides, past work, documents) for hyper-specific context.

3. Sharable Agents: Gems can be shared with colleagues, friends, or teams, massively accelerating team efficiency and standardization. You can share a standard Gem with your team, granting either Viewer access (to use only) or Editor access (to refine the Gem's instructions) for uniform, high-quality results.

WAIT, Just Before You Jump In… (Shine Ya Eyes)

1. Requires Clear Design: Just like every tool, it is only as good as the hand that wields it. The quality of the Gem depends entirely on the clarity and detail of your initial system-prompt instructions. If you get it wrong, it will too.

2. Platform Dependency: Currently, the full feature set is integrated deeply within the Gemini ecosystem, meaning you gats commit to the platform. Wahala for who no like Gemini before!

3. Lesser-Known Feature: Since it’s newer, the community templates and shared best practices are not as widespread as general prompting advice.

4. Instruction Sensitivity: If the initial instructions are contradictory or ambiguous, the output accuracy will suffer; garbage in, garbage out.

KOKO OF THE MATTER (Bottom Line)

Artificial Intelligence was invented to enhance human efficiency, but it isn’t efficiency if we spend hours battling with the technology. So, "I sabi use AI, I sabi use AI” no be by mouth! Is it really making you efficient?

With Gems (when used well), you don’t have to ask yourself this question. It mirrors everything a typical African AI user looks for: Resourcefulness, efficiency, and building something that lasts. Why waste time rebuilding the instruction set for your AI every single day when you can build a system that runs itself?

I challenge you this week: Don't just prompt Gemini — or any other AI. Spend one hour crafting a single, perfect Gem based on the most annoying, repetitive task in your professional or creative life.

You'll realize that the true power of AI is not in the conversation you have today, but in the systems you build to run tomorrow.