Diego Granados

I'm a Product Manager with experience working cross-functionally with business roles (e.g. Marketing, Finance, Sales), technical roles (e.g. Backend and front-end engineers, data scientists) as well as designers and UX researchers to bring products to life that make our users' problems and pain-points go away!

I have experience working with Web and Mobile products, AI and Machine Learning - There's something really fascinating about using data to try and predict the future. I'm a big believer that every model should be thoroughly thought from an AI Ethics point of view.

As a product manager I have to wear different hats every day and I excel at working cross-functionally. My experience and feedback has taught me that I bring value to the teams I work with because:

- I empathize with users and speak: geek, data whisperer, designer, marketer and salesperson. At some point I also tried to learn Shriiwook (I've always wanted to understand Chewbacca like Han did). Understanding means fewer mistakes, and mistakes can be very costly.

- I'm passionate about understanding customers, too many people use "customer obsession" as buzzwords in their resume.

- I leave the BIG ideas to the Elon Musks of the world. A good Strategy, Vision and Execution drives products forward - I do these three.

- I'm an engineer at heart and a musician at soul - I like to combine logic with creativity

- I'm the calm one. I'm the one that plans.

I'm passionate about helping others and I created a YouTube channel to help others learn about Product Management: https://www.youtube.com/c/PMDiegoGranados/

Don't hesitate to reach out if you want to talk about AI/ML, technology, business, consumer electronics/technology or need help on your journey to become a Product Manager!

Specialties: Product Management, Consumer electronics, Consumer Technology, Cross-functional work, Agile development, Product design, data driven decisions, AI, Machine Learning.

Product Manager AI&ML

I started AI-Prototyping my PRDs before I even sit down with my engineers.

Last week, I was working on a feature that was a bit of a nightmare in terms of complexity. It had a lot of moving parts for both users and admins. 

On paper, the PRD was solid—I had the requirements mapped out, the logic seemed right, and I felt good about it.

But because it was such a "heavy" feature, I wanted to make sure I could explain it to my team as clearly as possible. Besides, the number of comments and questions in the PRD was growing fast, and now an important part of the PRD context relied on someone reading multiple threads and comments

So, I took a screenshot of our UI, dropped it into Google AI Studio, and essentially prototyped my PRD with AI: 

💬 "From this screenshot, when I click on X, add a panel with information from this other screenshot".
💬 "Now, when someone clicks Y in the panel. If the user has permission A, then do Z, otherwise follow... (some other requirement in the PRD)". 
💬 "Add a button at the top of the product to toggle whether I'm an admin or a user. Admins can view X, Y, Z panel and button, but users cannot see them".

After a morning of working with AI Studio to convert my PRD into a working prototype, the "aha!" moment happened...

In my case, the user side looked great, just like I’d written it. 

But once I started "building" the Admin side... I realized I missed the mark. My plan for the Admin UX wasn't nearly as smooth as I thought it was on paper. It looked fine in the doc, but in a real flow, it was way too complex.

AI prototyping made things so much easier because I was able to: 
✅ Simplify the Admin UX: I saw the friction immediately and fixed the flow before the meeting even happened. 
✅ Catch Corner Cases: I found a few gaps for both users and admins that only became obvious once the experience was "in motion." 
✅ Ease the Handoff: When we finally had the review, we were looking at a functional reference, which made the discussion so much easier (and the PRD easier to clean now that we were on the same page)

I’m still using Gemini to help me write the initial PRDs, but I’ve now added AI Studio to my toolkit to help me "see" my ideas before I ship them to the team.

It’s helping me be a better PM and a much clearer communicator.

As features get more complex, a good PRD helps me share a good idea with the team... but now that I'm using AI Studio to see my idea come to life after a few hours and how easy is the conversation with the engineering (and design!) team, it has become a permanent part of my PRD workflow.

👋 Have you tried "running" your docs through an AI builder or no-code tool before? I'd love to hear if it's helped you catch things early! Drop a comment below.

-----
💎 AI Product Management can sometimes feel intimidating, so I put together a few resources for you in my comment below!

undefined undefined

Bento

👋 If you liked this post, you are going to LOVE  all the Product Management resources I have for you over here: https://bento.me/pmdiego

Product Manager AI&ML @ Google | 🚀 Interested in AI Product Management? Check my profile!

I saw a job posting for an AI PM at Figma yesterday, and it highlights why "vibe-launching" LLM products is not enough to become an AI PM.

Anyone can build an LLM-Wrapper over the weekend, but it's not enough to be an AI PM at companies like Figma, Google, Microsoft, Anthropic, and so on...

The reality is, this role was never just about prompting; it’s about owning the Machine Learning lifecycle.

I see a lot of aspiring AI PMs focus purely on the "creative" side of GenAI, but if you look closely at these job descriptions, they are asking for three very specific, very technical skills that define the role in 2026:

1. Beyond the "Black Box" (LLMs & ML Fundamentals)
Figma asks to "prioritize model improvements." You can't do that if you don't understand what's happening under the hood. For example:

🤖 LLMs (RAG vs. Fine-Tuning): 
If your chatbot fails, is it a Retrieval (RAG) issue (showed the wrong doc) or a Fine-Tuning issue (wrong tone)? If you don't know the difference, you can spend too much time 'fixing' the wrong thing.

📊 Traditional ML: 
Think about a Netflix Recommendation System. If it recommends movies you hate, it’s likely a data issue—maybe the model only trained on your weekend habits. You need to understand how Data Collection and Training work so you can spot these bias issues before they ruin the user experience.


2. Owning the "Definition of Good" (Evals & Metrics) 
In traditional software, a bug is a bug. In AI, "quality" is subjective—and that is terrifying for a roadmap. That’s why you see requirements for "experience with evaluation and iteration."

🥇 LLMs (Golden Datasets): 
You have to move beyond "it feels good". You need to learn how to build Golden Datasets—essentially a set of ground-truth examples that you define as the perfect answers. When engineering updates the model, you run it against this dataset. If the score drops, you don't launch.

🎯 Traditional ML (Context): 
You need to understand why an 80% Precision score might be great for a music recommendation, but 90% could be a total disaster for a fraud detection model. 


3. Scaling (Reliability & MLOps) Making a demo work for one person is easy. Scaling to 10,000 is hard.

When companies ask for "scaling experience," they are talking about the unsexy stuff: Latency, Cost, and Reliability. You need to get familiar with the MLOps landscape—tools like LangSmith or Arize for tracing errors, or Datadog for monitoring latency. 

----

The biggest hurdle isn't Python. It's moving from Deterministic code (If A, then B) to Probabilistic outcomes (If A, then probably B).

It changes how you think about roadmaps and how you manage user expectations when you can't guarantee a specific output 100% of the time.

👋 If you’re trying to move into an AI PM role, what's the biggest challenge you are facing?

---
💎 I’ve been an AI PM for 6+ years. If you want to dive deeper into AI Product Management, check my comment below for resources!

I’ve been an AI PM for 6 years. Back then, AI wasn’t "vibe-LLM" everything—it was mostly a lot of 'where do we find data?' and math.

When I was at Microsoft (pre-ChatGPT), being an AI PM meant focusing on data lifecycles and "black box" models. You spent your time explaining why a 90% accuracy rate was great for one thing, but a disaster for another. 

At Google, part of my job is to use Gemini to develop AI Agents that automate data science tasks for data scientists.

Looking back, I’ve realized "AI PM" isn’t just one type of job anymore. It has split into three paths:

1. The AI-Experiences PM: Focus on the "last mile"—how users interact with AI (like Amazon’s Review Summaries or Gmail’s Smart Reply). You don't need to code, but you must understand LLM Evaluations—measuring if outputs are helpful and safe before launch.

To break in, study AI UX patterns (how products handle hallucinations or latency). No-code tools like v0, Google AI Studio, or Lovable make it easy to prototype and test your ideas today.

2. The AI-Builder PM: Building the "engine" behind products like Gemini, ChatGPT, or Netflix’s recommendations. This requires technical depth: data quality, training pipelines, and metrics (Precision/Recall). You need to understand the "why" behind the data to know the model's limits.

I recommend formal education here—ideally a Master’s in CS or Data Science—to truly understand the infrastructure you’re building.

3. The AI-Enhanced PM: This isn't a title; it’s using AI to be a 10x PM. I recently used Google AI Studio to "run" my PRD logic before meeting engineers. Converting a doc into a prototype in two hours helped me find logic holes I’d missed and improved my engineering handoff. That is being "Enhanced"—using AI to be 10 times more productive as a PM.

-------

The biggest hurdle in the PM to AI PM transition is moving from Deterministic code (If A, then B) to Probabilistic outcomes (If A, then probably B). It changes how you manage roadmaps and user expectations when you can’t guarantee a specific output 100% of the time.

💡 Anyway, I’m curious—if you’re an AI PM, how did you end up in your role? If you’re trying to move into this space, what’s the biggest thing stopping you?

----

💎 I’ve been in this space for 6+ years. If you want to dive deeper, Marily Nika, Ph.D and I wrote "The AI Product Playbook." I’ll leave a link in the comments!

One of my favorite AI tools helps me 'chat' with a version of myself that remembers all the context about my product and everything I've worked on - Very helpful when I work on new Features, new PRDs, or Strategy docs.

I love using Gemini and AI Studio to create content and AI-prototype my PRDs... but NotebookLM has recently become one of my go-to AI products.

Like any other PM, I usually drown in 50+ tabs open—old PRDs, a collection of user interview transcripts, and raw data exports. 

Recently, I started playing more with NotebookLM, and now I chat with all the context I need - I love it!

The power behind NotebookLM is RAG (Retrieval-Augmented Generation). It creates a strict boundary around only what you upload. It forces the AI to "look up" the answer in your specific files rather than searching on internet.

This NotebookLM "Context" workflow has worked for me and saved me from overwhelming my browser (and me) with 50+ opened tabs:

1. Add Relevant Context: 
I have different "context" notebooks. 
For example, I have one for high-level research on my product and competitors:
📘 Context: I add user research, presentations that introduce the product, roadmap slides, strategy docs, industry research, and competitor research
⭐️ Why?: Very helpful for making sure that the product is addressing what our users are saying and how it compares against competitors.
💬 Example: "Help me find information about problems called out by customers related to X that we are missing from our roadmap. Do these features exist in our competitors, Y and Z?"
💡 Even starting with a simple notebook with all your user research reports is powerful enough! You can "talk" with your users and ask them questions!

2. Leverage Deep Research
One of my favorite features about NotebookLM is the ability to do 'Deep Research' about your users, your product, and your competitors.
📘 Deep Research Prompt: "I want to understand more about my users for X industry that use products like [my product], competitor A, B, and C. I want to see what users say about them in social media, articles, and industry reports".
💬 After you add the Deep Research results and sources to your notebook (combined with your own sources), you can prompt something like: "What do users like the most and the least about [my product] compared to competitors A, B, and C?" 
💡 Now you built a 'context vault' that gives you your information complemented with online research and clickeable citations!

3. Interact with your data
I use NotebookLM the most to "talk" to the context I need. However, I've seen a lot of crazy outputs online where people create infographics, presentations, 'podcasts' out of their data, etc.
💡 To me, the most important part is to understand that NotebookLM is as powerful as the data you put into it.

---
⭐️ What other AI tools are you using?

---
💎 AI Product Management can sometimes feel intimidating, so I put together a few resources for you in my comment below!

When I launched my first GenAI feature, I had to completely relearn how to define "Is it good enough to launch"?

I was comfortable with Traditional ML metrics. If you asked me about Precision, Recall, or F1 Scores, I could have a great discussion with Data Scientists about whether or not we are ready to launch.

But when my engineering lead asked me for the Go/No-Go criteria for our new LLM feature, those metrics didn't help me much.

He asked a simple question: "How do we know this is good enough to ship?"
To be honest? At first I didn't know what to answer...

I realized I could speak "Traditional ML Metrics", but I didn't know how to speak "LLM Quality".

My initial strategy was what most teams do: Vibe Check it and launch because there's pressure from an executive. 

We ran a few files through the model, read the outputs, and nodded. "Yeah, looks good. Ship it!" 🚀

That works for a prototype. (Don't do that in production...)

We tend to test for the 'Happy Paths' that we know our LLM can do, and tend to ignore thoroughly testing all the other things that might break your feature.

Through that launch (and a lot of research since), I learned that you need a mix of three specific evaluation layers to actually trust your launch:

⚡ 1. Code-Based Evals (Sanity Checks)
Start here - Use standard code (Python/Regex) to catch the "dumb" errors instantly like...
- Did the model return valid JSON?
- Is the answer under 50 words? 
It’s instant and free. It doesn't tell you if the answer is smart, but it tells you if it's broken.

🧠 2. Human Evals (Your Ground Truth) 
Most teams try to skip this layer because it is slow, expensive, and manual, and they want to jump straight to automation. It's a trap! 

You need real humans to grade the outputs to define what "Good" looks like. This creates your Golden Dataset—the source of truth that you measure everything else against. 

If you don't do this, your automated metrics are just measuring noise.

The toughest part about Human Evals is convincing your stakeholders (esp. leadership) that you need to spend ALL THAT TIME doing Human Evaluations - they are critical, don't skip them.

🤖 3. LLM-as-a-Judge (Fast and Scalable) 
Ideally, you'll use a stronger model to grade your own model’s output. Something like...
- Input: "Compare the Model's answer to the Golden Answer."
- Criteria: "Rate accuracy on a scale of 1-5." 

This lets you run thousands of tests in minutes. It allows you to scale your "human" logic without the human bottleneck and keep evaluating as you make changes.

The biggest mistake you can make is thinking you can automate trust.
You can't. You have to put in the hard, manual work (Human Evals) too.

---
👋 How are you currently measuring quality for your AI features? Are you using "vibes" or do you have a formal eval pipeline?

---
💎 Need help with Evals? George Zoto, Marily Nika, Ph.D and I put together a hands-on AI Evals course for you - Check my comment below for a code with a discount!

taplio

Diego Granados's Linkedin Analytics

Diego Granados

Performance ScoreDiego Granados's LinkedIn Performance

Who is engaging with Diego

Diego Granados's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Diego Granados's Linkedin Analytics

Diego Granados

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Diego Granados's LinkedIn Performance

Who is engaging with Diego

Diego Granados's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Diego Granados's Linkedin Analytics

Performance ScoreDiego Granados's LinkedIn Performance

Diego Granados's Best Posts (last 30 days)