Miguel de La Lama
AT&T
![](https://framerusercontent.com/images/FgE5KrrJlCd9BHA9vwJQYXed4.png)
![](https://framerusercontent.com/images/v9839Ohkr1fxg5fdlzHLt07qh4.png)
Miguel de La Lama is a Senior Product Manager at AT&T focusing on two products: personalization strategies for the main website, AT&T.com, and the website in Spanish.
Outside of that, he’s also creating different products for innovation that will eventually help grow the business. For example, a just-launched product uses their own LLM model and information they feed it to create copy for the website in a hyper-personalized way.
When you think about your workflows, and either the projects you're tasked with or what comes up over the course of a given quarter year, where do you usually start when there's information or a plan you have to build, and there's some information to gather, or research you might have to do?
First of all: is that true? And second of all, where are you starting if so?
For personalization, it's very open so I can touch any page, any business. And I always start as a consumer.
Like with my consumer app, I started by looking at our analytics. “Is it having any issues or what's happening?” I just need to understand how users are reacting. From there, it's like, “okay, we need to grow this” or “what's currently happening in the market?”
Then I usually start doing third party research.
Then, I take it all back to first party. It depends on data availability: does it exist? Or, sometimes the data doesn't per se exist for the industry, for the telecom industry, so I need to find different industries and just build a case.
When you think about the “data that exists” bucket, what are some of the challenges with collecting any of those “known knowns” or publicly- or internally-available sources of insight?
Third party is more about “Is it true?” What was the methodology? Was it a big enough sample? How did they collect the information? Was it biased, or can I use that information to build a case?”
So the validity of the data and research? Makes sense.
When you think about your overall workflow across the course of the year or otherwise, we've heard from a number of different companies and industries that there are three buckets of research and insights. I'm curious if this resonates and if one tends to align with you more than the others.
Big projects: We know something we have to do. We know the insights and the inputs to doing it well. And there's a clear output as well as a clear use case – what we're going to do with this information.
On the other end of the spectrum, there's ad hoc: We might get a question from a CEO, an idea from a colleague, or observe something we need to quickly get smart on, react to, or due diligence.
And then there's the in between: the always-on stay smart on the topic or stay relevant.
Do those three buckets tend to align? Do you tend to spend your time on one or the other?
I like all three if you ask me. For me, it's a ladder up.
The first bucket is definitely where or how we should be thinking because that's more related to the business and things that we need in order to move the right way.
Then the bucket about staying relevant – that's super important. The way I see it, all those three buckets are interconnected.
The least one that I might say is the ad hoc, because that is more subject to different goals. Or it might be that it can provide a piece of research, a piece of data, and that data might be misinterpreted or used in a different way that is not intended.
When you think about some of the experiences you've had with different external sources of information – whether that's analyst firms, Google, LLMs, or otherwise – if you could wave your wand and fix one of the problems with them (if there's any!), what would you fix or improve about the overall way of getting what you need from them?
That's a really good question. Answering the query, because I'm not just looking for the keywords.
And especially for me, with English as a second language, I might say different things – sometimes there’s a misunderstanding. That's why I like using LLMs as a tool to say, “Hey. This is my goal. Let's get it the best way.”
Currently I have been using a lot of for personal [use cases], because we cannot use Google Gemini. Gemini for search has definitely improved, and helps me with getting things faster and quicker.
You can't use Google?!
Google at AT&T? No. We have our own model from Microsoft, so we have our own version of ChatGPT through Azure.
Last couple questions! Easy speed round.
One word that comes to mind when you think about AI and research. How do you either feel about it or your optimism or not about it?
Connected.
When it comes to those Google / LLM type of solutions and the degree to which they meet your needs for research, is there an emotion or sentiment towards how they work?
They do help. Definitely needs to improve, but the technology is advancing super fast. So I'm really excited.
Super helpful. Thanks so much for the time!