Beyond the Prompts: Establish Your Framework to Use Generative AI Tools

Beyond the Prompts: Establish Your Framework to Use Generative AI Tools

Prompts matter and are essential to effectively use Generative AI (GenAI) tools such as ChatGPT, Claude, Gemini, Microsoft Copilot, Perplexity and others). However, what matters more is establishing a framework that allows you to efficiently maximize and optimize the way you use those GenAI tools and the effectiveness of the prompts.

Prompts should be the way you execute your framework, and not the framework in itself. A framework includes the working principles and establish the fundamental ways in which you will use a GenAI tool. The prompts execute those principles and ways.

Prompts will always be optimized for high speed, but low reliability or originality. This means that, unless you establish a clear framework from the beginning, when you prompt a GenAI tool you will get the same response, in milliseconds, than anybody else. However, with a framework, you will be creating the conditions for any GenAI insights to be adapted to you and your needs.

For example, if you work in HR and you want a GenAI tool to help you define a guiding policy for effective communication among managers and employees in fully remote organizations, you don't just prompt the tool with: "Design an effective communication policy for remote workers and their managers". That's what most users end up doing. And what you get as a result is a general, superficial, out-of-context, response.

But, if you first establish your framework, and then prompt the tool, you will get a better response. In addition, you will not have to come up with prompts or fixing existing prompts every time that you have a new request, but you can use the framework you established.

A Framework to Effectively Use GenAI Tools

These are the basic components of a framework. For the examples, I am using both Claude.ai and ChatGPT.

1 - Create your framework in a document outside of the GenAI tool

Your framework is a working document that you will improve over time. Often, the framework is independent of the actual information you need the GenAI tool to help you with at any given time. For example, today you may need to develop a policy for effective communication between managers and employees in a fully remote organization, but tomorrow you may need to develop a training program for first-time people leaders new to the organization. While the prompts may vary, the framework will continue to be useful.

Use any program (Microsoft Word, Google Docs, etc.) to write your framework.

2 - Ask the GenAI to act as the expert in INSERT HERE field

You can establish a working principle in your framework such as:

"For my request today, I want you to act as a communication expert, particularly in the field of remote and hybrid work, and virtual communications. Please respond to my requests as such an expert. Please don't do anything yet, just confirm that you understand that you will act as the expert I described".

3 - Start explaining the request and ask for clarification questions

It goes like this.

First iteration: Initial Request

"I need you to help me write a communication policy that supports direct managers and employees in my organization. Our organization is a fully remote accounting firm with 150 employees located in various cities in the United States. The communication policy must establish the guidelines for the way in which our managers and employees will interact. This includes feedback sessions, performance management, ongoing requests, out-of-work hours communications, priorities and urgencies, etc. I will share with you a few policies for you to understand our writing style. Before that, please confirm that you understand my request. Also, you have permission to ask me clarification questions. Please begin by asking me 10 questions related to this request. Please don't start developing the policy yet until I have addressed your questions and give you the confirmation that we can start creating it".

Second iteration: Request, including responses to GenAI questions (you don't have to respond to all the questions, and also you can ask for more clarification questions)

"I need you to help me write a communication policy that supports direct managers and employees in my organization. Our organization is a fully remote accounting firm with 150 employees located in various cities in the United States. The communication policy must establish the guidelines for the way in which our managers and employees will interact. This includes feedback sessions, performance management, ongoing requests, out-of-work hours communications, priorities and urgencies, etc. I will share with you a few policies for you to understand our writing style. Before that, please confirm that you understand my request.

Some of the challenges and pain points regarding communication between managers and employees include the fact that communication is too formal, generally done via email, responses take a long time to come back, managers reach out to their teams out of work hours. We currently don't have any policy in place, so this will be the new one. We would like for managers and team to use other real time tools such as Slack, Whatsapp, or even text messages. We want to change our culture and encourage managers not to reach out to their teams outside of work hours. Particularly, we will penalize them if they do unless it is a real emergency. Time-sensitive issues must be prioritized and employees will always, by default, have permission to halt their current workload to address these time-sensitive urgencies. The policy will apply to all levels in the organization and we want to implement it within the next two months.

Please, don't write the policy yet. Before proceeding, start by confirming that you understand the specifics of this request. In addition, my next prompt will include our writing style so that you know how to write our policy".

4 - Provide the example of your (personal or organizational) writing style

The first lines of your framework can be something like this:

"These are several links (or: below see the text of) to existing corporate policies in my organization. Please analyze the content so that you understand the writing style in which my organization establishes its corporate policies, particularly employee policies. Please don't do anything yet, just confirm that you have analyzed the writing style".

This will allow any future prompt related to similar requests to respond in a voice and style that match that of your organization.

However, this would be an important addition to the paragraph above:

"In addition to following our voice and writing style, please do not include the common fillers or words that you commonly use, such as: in the landscape of...., or the words nuanced, fabric, weave, necessitates, etc. You must focus on our writing style, but also be creative with the use of words".

(we are not including an example of this as this may vary significantly from organization to organization).

5 - Break It Down (ask GenAI to break it down for you)

"What are the sections that you think a policy like this may include. Please take into account all information provided before for context".

6 - Start

"Using the context provided before, the writing style, and all the information discussed, please write the content for the first section of your suggested structure"


Not Perfect... But Good Enough

GenAI responses don't have to be perfect. And they will not be. However, if by using any of these tools you can have a 20% headstart (compared to not using them) and save 50% of the time in the rest, then you will find yourself in a place of significant added gains and efficiencies.

Someone said once: "AI will not replace your job. Someone using AI will". I don't know if this will hold true forever, but, at least for now, GenAI does help. If writing a communication policy from scratch, pre-GenAI would have taken you one week, and now, with GenAI, it takes you two days, then that is a significant optimization.

Finally, you can create your free accounts for Claude, GenAI, Copilot, Perplexity. There are other models and tools you can use.

One last note: please don't share private or proprietary data or content in any of these platforms.


By Enrique Rubio (he/him)

(Note: I write fast, speak fast and read fast. Sometimes - often times - I don't see the typos! Please let me know if there are typos or grammatical misconstructions in my write ups. It happened to Shakespeare, it certainly can happen to me. Do so with kindness, compassion and grace, though. I appreciate it!)


ENROLL TODAY: Mastering ChatGPT (For HR Professionals)

Introducing the first, ever, LIVE Certificate Program focused on mastering ChatGPT for strategic and tactical HR.

The "Mastering ChatGPT (For HR Professionals)" mini Certificate Program (6-weeks) is strategically designed to equip HR generalists, specialists, and leaders with the knowledge and skills needed to integrate advanced AI technologies, like ChatGPT, into their HR practices effectively.

Spanning the essentials of AI in HR, tactical and strategic applications of ChatGPT, and innovative HR practices, this curriculum ensures participants are well-equipped to enhance recruitment, onboarding, culture building, and more through AI.

Enroll today.


Download Our Full Guide

Click here to download.


Hacking HR

We are powering the future of HR!

Hacking HR is the fastest-growing global community of people leaders and professionals interested in all things at the intersection of people, organizations, innovation, transformation, workplace and workforce, and more. We deliver value through hundreds of events a year, community engagement opportunities, learning programs, (soon) our certificate programs, and more. Join our community platform, the Hacking HR LAB. Click here.


Sponsoring the Hacking HR's Newsletter

  • Hacking HR is one of the largest HR communities on LinkedIn and the number one global community in terms of engagement.
  • We have over one million community members across all our platforms. Our LinkedIn page has almost 650k followers, more than 60k members in our Hacking HR LAB community platform, and over 434k subscribers to our LinkedIn newsletter (the largest and most engaged HR newsletter in the world!).
  • Click here to find out about our sponsoring plans.
  • The ROI for sponsoring our events is much higher than any other existing HR community, and it is also at a much lower price. If you are interested in sponsoring this newsletter with content-oriented subject matter expertise and thought leadership, please reach out to us at [email protected]


Fascinating read, we're also exploring the power of prompting especially in the context of image generation and it's pretty fascinating how much the end result is influenced by prompting. ?

回复

Thanks for sharing

回复
Rashmir Balasubramaniam

Enabling leaders with big missions to unleash potential in themselves, their people and their places

5 个月

Love the request for 10 (clarifying) questions. That is a useful thing for humans and AI alike!

回复

要查看或添加评论,请登录

社区洞察

其他会员也浏览了