In December 2023, they created a self-assessment - measuring engagement and introducing skill level.

We asked them to show us their conversational prompts. We asked them to show us their quality control prompts, give us an example of the work they could do.

They’re going through a process called top-grading, which means you’re below the bar for the role. They want you to hit the bar and will spend time helping you get there. If you don’t, you’ll be replaced.

It is 100% our goal to not replace anybody. It is to get everybody up to snub, bring everybody up to the same level. And this isn't like we're not trying to be the experts. We're just trying to be like base level.

So AI Mondays were introduced for 2024, without equivocation.

You may not talk to a customer, or have a video call. Go visit a customer, go to a trade show or do any of the things you normally do. You can't do anything except work on Gen AI projects inside the company.

They hired a VP of Gen AI who spent a decade on ML at Ericsson to develop a curriculum for learning various techniques and capabilities.

Now everyone has only 4 days to do what they used to do in 5.

But that’s by design, you’ve only got 4 days to do the same work, and the way to do it is by leaning into using AI.

Everyone received $1,000 for AI education and were encouraged to spend it on 'AI Mondays’.

From the self-assessments, they realised that over 50% of people were using very basic approaches to using AI.

We realized you can't ask somebody to build a house until you teach them what a hammer is, what a screwdriver is, what a saw is... We were doing that. We're like, 'Go build a house. Let's see what your house is.' So we've spent a lot of time on prompt frameworks, for example, and best practices.

So IgniteTech has focused on educating it’s team by:

Equipping people with the basics

Helping people understand; what is a prompt? What’s the best practice for a prompt? Don’t use the little prompt input box as it’s inherently limiting—go create a prompt in a Google Doc with delimiters and steps clearly outlined. Try that prompt with different LLMs; OpenAI, Claude, Bard (now called Gemini). Ask the model to rewrite it based on best practices. Ask it to reference people that you want it to model after.

Prompt frameworks

Borrowing from experts, they leaned on Sheila Tao’s CO-STAR prompt framework, for which she won Singapore’s GPT-4 Prompt Engineering Competition. Here’s how it works:

(C) Context: Provide background information on the task

This helps the LLM understand the specific scenario being discussed, ensuring its response is relevant.

(O) Objective: Define what the task is that you want the LLM to perform

Being clear about your objective helps the LLM to focus its response on meeting that specific goal.

(S) Style: Specify the writing style you want the LLM to use

This could be a particular famous person’s style of writing, or a particular expert in a profession, like a business analyst expert or CEO. This guides the LLM to respond with the manner and choice of words aligned with your needs.

(T) Tone: Set the attitude of the response 

This ensures the LLM’s response resonates with the intended sentiment or emotional context required. Examples are formal, humorous, empathetic, among others.

(A) Audience: Identify who the response is intended for

Tailoring the LLM’s response to an audience, such as experts in a field, beginners, children, and so on, ensures that it is appropriate and understandable in your required context.

(R) Response: Provide the response format

This ensures that the LLM outputs in the exact format that you require for downstream tasks. Examples include a list, a JSON, a professional report, and so on. For most LLM applications which work on the LLM responses programmatically for downstream manipulations, a JSON output format would be ideal.

Source

Hands-on practice

Give people a real use case to put it into practice.

everybody had 75 minutes to complete this exercise. They didn't know it was coming. It was a timed exercise and they had to choose one of two assignments. One was to write or to come up with AI features for a set of products that they had never heard of, for a company they had never heard of.

And the other was, imagine that we've acquired that company because we are a very acquisitive, company. Design the restructuring for HR and all the processes in the company. How will you measure it?

Now, a year ago, people would have said, you know what? The first thing they would have done? Is open Google and gone and found the company and started reading about the products. Yeah. You're not going to make it through this in 75 minutes.

They require participation, but mostly focus on mindset over skills.

The training seems to be paying off in expanding employees' perspectives. As Vaughan noted after a recent AI Monday training,

I got some beautiful answers, but I got people that not only did they design features for these four products that we listed, but they even had it right press releases announcing the new features. Some people were flexing their muscles. Like I can do this stuff now.

Iterative learning

Give groups the same assignment to complete individually and then come together and compare their prompts. Select the best one and enhance it.

Improving products with AI

IgniteTech is also undertaking a sweeping effort to incorporate AI across its software offerings and internal processes.

We are now going through every single process. Applying what we know right now as best practices in Gen AI, brought from months of working on this to completely redesign those processes to eliminate manual work and stopping waiting on a person wherever we possibly can.

The most fundamental thing is how you start the day. And you are about to do something. Your very first thought should be, can I do that differently? Can I include a tool to augment that process? Is there something I could try? That's what AI first is.

The company announced plans to roll out AI features in each product division. This requires a significant investment but promises to make Ignite Tech's solutions smarter, more efficient, and more valuable.

We announced to the world, all of our customers, that we will have an AI capability in all of our...we've articulated divisions for each of them, and we're starting to roll all of those out.

They are not only rolling out AI in all their products but have found new areas they can develop.

They acquired three companies under one product, GFI, but it was operating separately instead of as one.

Certain resellers only think of one of them, never talking about the other, no cross-training, etc.

So they built App Manager, bringing the three interfaces under one roof.

Now all three products were under one roof, and the data was unified.

They were able to see insights and anomaly detection built into a new feature called Radar, which can correlate data across products by implementing AI.

Leadership commitment is critical for AI success

Implementing AI on an enterprise scale requires leadership buy-in and commitment.

Eric made AI a personal priority at every level of the organization, and this leading-by-example approach is essential.

I've got to be involved as everybody else, and you've got to make this a priority. I think they should make it a priority, because if not, then in my opinion, we're going to quickly create two classes of companies and workers. Haves and have nots.

Other examples of companies leading AI adoption from the top can be seen here.

Give everyone the same chance to participate.

So in my opinion, as a leader of an organization, when you're trying to affect cultural change you've got to give everybody the same chance. The same support and the same encouragement to participate. But you got to find the ones that are like, yeah, I'm just not going to do it.

If you found this post valuable, share it with a friend, and consider subscribing. Feel free to suggest new topics for me to cover.

Cheers,