Using artificial intelligence to write Amadeus’ OKRs – Sanremo 2024

I asked ChatGPT and Gemini to help me write the Amadeus OKRs. What came out is a guide on the use of AI in strategy, the pros and cons of the two AIs, and which one to choose to be supported as a leader in strategy work.

Disclaimer: What I am about to tell you is pure speculation. I have no contact with the leadership of San Remo, the strategy and the numbers are made up, but the process I used can be replicated in your company for which I am sure you will have more accurate information.

StrategIA

One of the fields of study I am investing a lot of energy in is the use of artificial intelligence in strategic work and corporate leadership roles. If you are passionate about these topics, I also invite you to read:

1. The CEO of the future will be an AI
2. The C-Suite of the future (part 1)
3. The C-Suite of the future (part 2)
4. The first AI-Coach of STRTGY

I think Strategy is nothing more than the algorithm by which we enable everyone in the company to make the right decision. And while in the past algorithm meant a sequence of concatenated logical conditions (e.g., if) that made the concept too rigid to have practical application. Today everything has changed.

Artificial intelligence is capable of relating information with a speed and accuracy impossible for humans and consequently adapting the next step based on the newly acquired information.

Fascinating scenarios open up before us in the construction of business strategies and their execution.

This article is an example of that because the quality of the strategic information I will show you-though without any real foundation as I told you in the disclaimer- far exceeds the standard to which companies work today.

All of this, in relation to the cost and speed with which I got them, urgently calls for reflection on how to reconfigure the role and tools of current leadership and consequently also compensation.

Using the prompts I will provide you will be able to do the same from two interesting perspectives:

1. Using AI for your company’s strategy
2. Use AI to analyze and anticipate your competitors’ strategy

How to go from strategy to OKRs with AI

Step 1. Obtain the strategy

It is not possible to write good OKRs without knowing the strategy so the first step was to figure out how to get the strategy of San Remo using public information.

I wondered who, where, and when might have communicated it. I also believe that communicating the strategy is an excellent marketing opportunity that is rarely considered.

The densest document of this kind I have found is the video of the Feb. 5 opening press conference available for streaming here. All of the top leadership communicated their initiatives and celebrated the results with reporters. Obviously this is not “the strategy” but the effect of the strategy so, by reverse engineering, one can identify what the business objectives that drove the work might have been.

Unfortunately, the 88-minute streaming video is protected for download and I could not obtain the transcript to provide to AI but I know that the journalists present would have produced the material I needed. I identified 3 articles with which I ran and refined the process.

1. The Press
2. Evening Courier
3. Fanpage

The first article is a summary while the other two are live coverage that minute by minute go over the highlights. I ran the entire process with all three articles and the best result I report here was with the one from Fanpage.

Need to know. The quality with which the strategy is communicated is related to the quality of its execution. In the OKR Toolkit at MAKE PROGRESS you can use the Strategy Book which gives you a simple and effective outline for communicating all aspects of strategy and getting input from teams. You can find it in the book or purchase it separately.

ChatGPT

Here was the first major difference between the two IAs.

ChatGPT said several times that he could not access the page and that, knowing the event he could help me anyway. In the first few tests I provided him directly with the texts of the articles by copy and paste. After a few tries instead he was able to read the article independently. The answer is simple.

Gemini

Gemini, on the other hand, had no trouble reading and processing the content. His response is much more articulate. He seems to have understood the strategy better and describes it as an executive summary in a very readable bullet point. It was also noticeably faster.

Step 2. Writing OKRs.

After understanding the strategy, I used this simple prompt to instruct both AIs to write OKRs following the format of MAKE PROGRESS, which allows them to always and accurately intercept the business priority.

Need to know. It is important to provide the AI with a structure with which to deliver the output in order to achieve the best possible results and most importantly consistently. In our trainings we educate leaders to do the same in real life because OKRs form the grammar for communicating the strategy so it is important that everyone uses the same rules to do so. The strategy will then be cohesive and consistent allowing everyone to understand it and offer their input accordingly.

ChatGPT

The nuances between the two platforms can be seen in the results of this task.

Here is ChatGPT’s result highlighting 3 priorities: visibility, engagement, business partnerships. As always, it is precise and concise.

Gemini

Here are Gemini’s results below, and I must say he seems to have had a better understanding of the strategy. The priorities it identifies are 3: audience, revenue and impact for Italian music. Particularly the latter I would not have expected from an AI!

I also appreciated the paragraph in which he connected Amadeus’ goals with business goals.

Step 3. Connect the OKRs to the execution

Writing OKRs does not ensure that the goals will be achieved, which is why it is important to coordinate the execution part.

To do this, I used a technique described in MAKE PROGRESS by asking the AI to identify Leading metrics, i.e., metrics that are immediately actionable by teams and anticipate the achievement of the goal itself. For each metric, I asked it to propose 3 tasks and identify the responsible team. Here is the prompt.

As you can see, the prompt is very simple and discursive for two reasons.

The first is that my account has Custom Instruction that prevents me from repeating pieces of prompts that would otherwise be redundant and make the conversation cumbersome. For this reason if you use my prompts without customizations you might get different results.

The second is that I don’t think very long, structured prompts bring better results. Kind of like with people, if you take too long to explain something you probably don’t have a clear idea of how to do it and you’re putting too many constraints on tapping the other side’s creative potential.

After some testing, and as you see from these results, I have found a trade-off that I am happy with. I see the AI output as a starting point, a way to not start from the blank page, certainly not a final product.

Need to know. The Amadeus KRs are lagging metrics and therefore will only move more prominently in the final phase of the festival; some will need to wait even longer. It is common for leaders to have these kinds of metrics being accountable for long-term goals. For teams, on the other hand, it is advisable to have metrics that move as fast as possible, measured on a daily or weekly basis, in order to swerve quickly. At MAKE PROGRESS there is a very powerful exercise to derive leading metrics from any process, it is called KPI Chain.

ChatGPT

Here are the answers from ChatGPT, which proposed some very precise metrics and activities.

Gemini

Here is Gemini’s answer. Equally precise and surprising as a quality of thought.

Here Gemini had some technical problems.

While ChatGPT seems to be able to cut its responses to a size it can handle, Gemini freezes even though it has not finished giving the response. I was forced to type “continue” to continue, and unfortunately some answers remained incomplete.

In any case, it is a pity because the directions are surprisingly good.

Here are the next answers.

Which AI to use for strategy?

In this particular case ChatGPT seems to be more hermetic and less creative than Gemini, which surprised me because it captured nuances of the seemingly more refined strategies.

Gemini seems to have an edge on this one, too bad for the technical problems that make it not as reliable as its competitor.

There is also to point out that while I used ChatGPT 4 in a paid account, Gemini was in its free version.

The discriminator is not the AI to be used nor the budget but the quality of the information provided, and this is usually the most important hurdle to overcome.

As you can see, if I had fed the AI a budget or a budget, we would have achieved the same effect that we systematically get in business: disinterest in the strategy, difficulty in understanding and consequently in execution.

With these prompts you can get the basis for a strategy in a matter of minutes, without investing in expensive, misleading and inapplicable advice.

These tools enable leadership to increase its strategic acumen. Deciding not to take advantage of this opportunity means that you have deliberately chosen to limit the company’s growth and jeopardize execution.

Let me hear from you if you use the tips I have recounted in this guide, I care.

ALWAYS MAKE PROGRESS ⤴
Antonio