CHATGPT, Google Gemini, How to Write

How to Create a ‘Bespoke’ Training Plan using Google Gemini

Posted on by Anthony James
11
Jun

One of the most practical applications of AI is its ability to create ‘bespoke’ training plans. The limitation of most training courses is that they are too generic, cover material you already know, and fail to provide the level of detail you want.

In this tutorial, I’m going to walk you through how to create a tailored training plan. To give it some context, I’ll develop a plan to upskill a remote IT team. I’m using this as an example as increasingly we’re working as part of remote teams, and I suspect that coordinating the onboarding, upskilling, and development for remote workers can be quite challenging. While there are a lot of generic training materials out there, how do you find a way to carve out a bespoke plan for your team?

If you’re responsible for developing a training plan for a remote, distributed team, it can be a struggle to know where to start. While you can use a training plan template to get you up and running, working out the actual modules can be a bit of a pain. Where to start? What level of detail do you need? How do you assess whether the plan is effective?

This is where AI comes in. Personally, I use Google Gemini, but you can adapt the following to most other LLMs, such as ChatGPT, Claude, and Hugging Face.

Let’s look at AI, in this instance Google Gemini, and see how we can use it as a training consultant to develop a bespoke training plan. If you haven’t used Gemini before, I’d recommend reading some of the other tutorials to come up to speed. In this tutorial, I’ll use the standard, free version, as this is what most of us are on.

Here’s the approach we’re going to take:

Download Training Plan templates, forms and spreadsheets – MS Office

#1 Clarify Your Expectations of Google Gemini

Explain to Google Gemini exactly what you’re expecting.

“Do I need to explain everything to you?”

Ever heard this in the office? We’ve all been in a situation where one person assumed the other would grasp what they were trying to do – and they didn’t. The curse of knowledge is frustrating for both parties. And this is where most people go wrong with AI.

In order for it to provide you with bespoke training collateral, start by asking it about its preferences in terms of collaboration. In other words, what’s the optimum way to get results from the AI? Instead of using generic prompts, we’ll look at how to structure the prompts so you get the information you need.

Q: What is a prompt?

A: In the context of AI, a prompt is an instruction. When you give the AI a prompt, you’re asking it to provide a specific answer in response to a specific set of instructions. See this: Google Gemini FAQ on Prompts.

The simplest way to get started is to enter the following prompt:

“My goal is to create a detailed training plan for a 10-person remote IT software development team. As a training expert with access to multiple training plans, can you suggest the optimum way for us to create this plan? Ideally, I want a plan that will allow me, an IT Development Manager, to train, test, and evaluate my staff. Can you outline a suggested approach to do this?”

It then thinks about this but… the response from Gemini isn’t really what I’m after. It starts to tell me how to create the actual plan, not how we can work together. Here’s a snippet of the response: 

Step 1: Needs Assessment and Team Evaluation

  • Identify Skill Gaps: Conduct individual or group interviews to understand current skill levels and gaps in specific areas (programming languages, frameworks, methodologies). Consider past project challenges and desired future projects.
  • Team Roles and Learning Styles: Identify individual roles within the team (front-end, back-end, DevOps) and preferred learning styles (visual, auditory, kinesthetic). This will help tailor training methods.”

As you can see, this is fine but hasn’t met our expectations. Let’s look at how to fix this.

#2 Use Iterative Questions and Answers to Collaborate with Google Gemini

Here’s something to consider when using AI. Instead of expecting it to give you the correct, definitive answer straight away, use a question-and-answer format to progressively deepen your understanding of the topic. For instance, imagine you want to know more about Superposition. You want to truly understand how it works. A one-sentence definition isn’t going to help. Instead, you’ll need to grasp the fundamentals, then iteratively go into more details. See how Sheroy Cooper explains it on Medium.

For our remote team training plan, this means we’ll go through an iterative process to define the shape, contents, and evaluation methods. To remedy this, I’ll change the prompt I entered above as follows:

“Goal – Create a detailed training plan for a 10-person remote IT software development team. As an IT training expert, suggest the optimum way for us to collaborate to create this plan. Discuss what you expect me, an IT Development Manager, to provide you with so we can develop this plan.”

Note: I explicitly said ‘IT training expert’ to give it more context. Remember, its database has hundreds of thousands of fitness training plans. Make sure to exclude these from your results.

This response is now closer to what we’re after.

Gemini says: “I’ll need details about the current skill sets of your team members. This could include a breakdown of roles, their experience levels with different technologies, and any identified skill gaps.”

A ‘conversation’ has now started. It’s prompting you to give it more context, background information, examples – anything that will help it prepare the training material.

Here’s the first part of its response:

Optimum Collaboration for a Remote IT Training Plan

To create a detailed and effective training plan for your 10-person remote IT software development team, we can collaborate in the following ways:

  1. Information Gathering:
  • Team Roles and Skills: I’ll need details about the current skill sets of your team members. This could include a breakdown of roles (developers, testers, DevOps engineers, etc.), their experience levels with different technologies, and any identified skill gaps.
  • Training Needs Assessment: Understanding your team’s specific needs is crucial. What are your upcoming projects? Are there new technologies you want to adopt? Are there areas where your team struggles with efficiency or collaboration?
  1. Planning and Development:
  • Training Goals: Once we understand the needs, we can define clear training goals. These goals should be specific, measurable, achievable, relevant, and time-bound (SMART).
  • Training Format Options: Given the remote nature of your team, we can explore various training formats, including online courses, live virtual workshops, self-paced learning materials, or a combination. I can suggest resources and platforms based on your budget and preferences.
  • Learning Assessment: I can help design methods to assess learning outcomes. This could include pre- and post-training assessments, knowledge checks within online modules, or project-based assessments.
  1. Implementation and Support:
  • Delivery Schedule: We can create a timeline and schedule for the training plan, taking into account team availability, project deadlines, and the complexity of training materials.
  • Remote Training Best Practices: I can share best practices for facilitating remote learning, such as using interactive platforms, encouraging knowledge-sharing sessions within the team, and ensuring clear communication throughout the process.

Here’s the interesting part. At the end of the response, Gemini lists exactly what it expects of us:

Gemini response

“What You Can Provide (as the IT Development Manager):

  1. Team Information: Provide a breakdown of your team’s roles and skillsets.
  2. Training Needs: Identify your team’s specific training needs based on upcoming projects, new technologies, or areas for improvement.
  3. Budget and Resources: Share your allocated budget for training and any existing learning resources within the team or organization.
  4. Team Availability: Outline potential time constraints or limitations on your team’s availability for training.”

So, let’s answer one at a time. This is YOUR response.

I suggest you answer each of these one at a time. For example, say:

“Prompt: Here is the requested information for 1. Team Information. Here is a breakdown of my team’s roles and skillsets. I’m now going to give you details about the current skill sets of my team members. This includes a breakdown of roles, experience levels with different technologies, and any identified skill gaps.”

Next, attach the CVs and skills matrix.

If you’re using Claude, you could bundle up all the CVs into a zip file and upload it. With ChatGPT Premium you can do the same. With Gemini, feed it a single MS Word document which includes all of their CVs. Include as much detail as possible. Note: Before doing so, remember to remove any personal data, such as names, emails, etc.

“Prompt: Provide me with a training matrix which lists the skills my team needs to learn. My team includes junior and senior developers, testers, and SRE. Remember, they are all working remotely, so we need to use remote training tools to teach them. We also want to train them to use remote tools, such as Teams, Slack, and other popular cloud-based tools.”

Gemini provides a matrix which you can either copy/paste or open in Google Sheets. Here is an example of what the output looks like when formatted for readability. The export option is one of the strengths of Gemini. If you work in the Google ecosystem, I’d recommend exploring this to see how you can refine this process.

I suspect there are other ways you can connect to other Google tools, such as Gmail as well.

As mentioned above, the aim really is to keep asking it increasingly detailed questions, adjust your prompt, and export the material that meets your requirements.

Before I forget! You can also use this to develop a personal bespoke training plan for yourself. I’ve done this to help me get back into swimming. As always, the more detail you provide it with, the more specific the response.

#3 Export Responses to Different File Formats

Finally, we’ll ask it to prepare the training material in your preferred format. Depending on your AI tool of choice, you can export material to text, CSV, markdown, even UML images for flowcharts. In the previous section, we ‘exported’ the files to Google Sheets. Actually, technically we didn’t. Rather, it opened the contents in Sheets for us, which might seem like we’re splitting hairs but…

To find out what formats you can export to, I asked Gemini: 

“What file formats can you export to?”

Gemini said…

“I can’t directly export information to any file formats. As a large language model, I am designed to interact and generate text. However, I can provide the information you need in a variety of formats through my responses. For instance, the training matrix I created for you can be easily copied and pasted into a spreadsheet application (like Excel or Google Sheets) for further editing and customization.”

In other words, it can’t export to .CSV or .DOC but it can open applications that support the content. What’s important here is how we phrase the question. So, I tried again:

“Can you export to HTML?”

“I can’t directly export information to HTML format like I can’t (typo) with a spreadsheet. However, I can generate the HTML code for a basic table that represents the training matrix we discussed. You can then copy and paste this code into an HTML file or use it as a starting point to create a more styled table.”

Here’s the HTML code for the training matrix:

I made the mistake of asking it if it could export to a format. It took this literally. Technically, it can’t export to CSV, for example, but it can open it for you in Google Sheets. Also, it can’t export to an HTML file but it can generate HTML code. See the difference? This is why it’s important to play around with different AI tools to develop a sense of how it interprets your requests.

“Prompt: Can you save to a diagramming tool format?”

“Response: I can’t directly save the information into a format specifically readable by diagramming tools. However, I can help you translate the training matrix into a format that can be easily imported into many diagramming applications. Here are a couple of options:

  1. Text with Separators: I can export the training matrix as a tab-delimited text file. This format uses a tab character to separate each data point in the table. You can then import this file into your diagramming tool and define the separators to create a table within the application.
  2. Mind Map Structure: I can rearrange the information in the training matrix into a mind map structure. This can be helpful if you want to visually represent the relationships between different skills and roles. You can then copy and paste this structure into your diagramming tool and customize it further.”

“Prompt: Can you save to UML?”

“Response: I can help you translate the training matrix into a UML class diagram that partially captures the relationships within the plan.

classDiagram

class Role {

    name

}

class Skill {

    name

    level

}

class TrainingTool {

    name

    type

}

Role <-> |has many| Skill

Skill <-> |can be taught using| TrainingTool

Using Gemini, you can save your content into:

– Text

– CSV

– HTML

– Markdown

– UML Classes

If you have the premium version of ChatGPT, you can use the diagramming GPTs to automatically create diagrams on the fly. It’s quite impressive. Worth getting a short subscription to try this out.

#4 Workflow

One of the recent breakthroughs with AI tools, such as ChatGPT and Gemini, is that you can connect your chats to SharePoint, Google Docs, and Sheets. Using this approach, you can start to create a simple workflow between Gemini and your training repo projects.

The Finer Points

Instead of using AI (ChatGPT, Gemini, Bing) as sophisticated search engines, look for use cases where you can build unique digital assets. In this example, we explored how to create a bespoke training plan for a remote IT department. The three immediate benefits of this approach are that you can: 1) customize complex training plans, 2) tailor role-specific content to individual roles, and 3) update the curriculum based on evolving team requirements.

Finally, this tutorial shows how to define specific bespoke material. If you use the format we’ve suggested above, you can start to develop just-in-time training plans for your team, reduce your training budget, and evaluate your staff’s progress more effectively.

Learn More:

Anthony James