These workflows desperately need an AI refresh, says Global Product Leader
Discover which product management workflows benefit most from AI, and how leading teams use it to accelerate discovery, improve decisions, and increase learning velocity.
Tony Pagliocco
Associate Director of Global Product Management at Aldar
AI Integration: AI is being embedded into product workflows at Aldar, transforming decision-making and development processes.
Data Fundamentals: Effective AI application requires strong data foundations; poor infrastructure reveals flaws, limiting AI impact.
Discovery Redesign: AI is streamlining product discovery by synthesizing vast data quickly, reshaping traditional, manual approaches.
Human Judgment: AI aids product activities, but human insight remains crucial for interpreting context and making key decisions.
Adoption Challenge: Embedding AI into daily operations improves adoption, shifting team mindsets from resistance to practical use.
Tony Pagliocco is the Associate Director of Global Product Management at Aldar, where he is responsible for product strategy, operating frameworks, and ways of working across 24 product-led pods.
He and his team are focused on baking AI directly into their product workflows. So, we sat down with him to learn which workflows he's currently optimizing and how he's ensuring adoption across the organization. Here's what he said.
Complex problems at scale
I’ve spent the last 20+ years building and scaling digital products across industries, from media and gaming to aviation, logistics, and now real estate here in the UAE. I consistently solve complex problems at scale and build teams that deliver impact.
Want more from The CPO Club?
Sign up for a free membership to complete reading this article:
Have an account? Log In
Earlier in my career, I was very hands-on — launching products and learning to execute. I then led larger product organizations and drove enterprise-level transformation. One of the most formative chapters was at Boeing, where I led product for digital smart factory platforms supporting the 777 and 787 programs. We worked with multi-petabyte data environments, applying NLP and computer vision to manufacturing and supply chain workflows. We built systems to predict issues, optimize operations, and operate at global scale. That experience shaped how I think about data, platforms, and now AI.
This moment is an inflection point. We’re moving from building software to building intelligent systems — and the role of product leadership is shifting from managing roadmaps to orchestrating data, AI, and real-world outcomes at scale.
At Aldar, AI is deeply embedded into what we build, from smart city platforms and digital twins to master planning and feasibility tools. We use AI to simulate, predict, and optimize decisions long before anything is physically built, which fundamentally changes how the business operates.
This moment is an inflection point. We’re moving from building software to building intelligent systems — and the role of product leadership is shifting from managing roadmaps to orchestrating data, AI, and real-world outcomes at scale.
Leading product strategy
Today, I serve as Associate Director of Global Product Management at Aldar, reporting to our Chief Digital Officer, where I lead product strategy, operating frameworks, and ways of working across 24 product-led pods.
Our teams span a broad ecosystem, including property sales, retail, commercial, residential, loyalty and marketing, e-commerce, hospitality, and smart city IoT — and our flagship platform, Live Aldar, underpins all of it. We’re supporting both customer-facing experiences and internal platforms, with a strong focus on connecting everything into a seamless, end-to-end journey.
For delivery, we operate through cross-functional pods aligned to business outcomes, while investing in shared platforms like identity, payments, data, and AI to scale across the ecosystem. My role is to ensure consistency in how we build, while enabling each domain to move fast and deliver impact.
Why AI requires good fundamentals
Here's the first thing that needs to be said: AI is not a feature. It’s a capability that changes how your entire product and organization operate.
Many teams approach AI like any other enhancement — adding a chatbot, automating a workflow, or improving a feature. That mindset limits the impact. The real shift is that AI changes how decisions are made, how products are built, and how value is created.
It also exposes the fundamentals. If your data is fragmented, your workflows are unclear, or your product thinking is weak, AI will amplify those problems. Teams that struggle with AI usually try to layer it on top of broken foundations.
The real work is ensuring it has the right inputs, structure, and guardrails. Without these, outputs may look good on the surface but lack reliability for real product use. If I’d known that upfront, I would have invested earlier in data cleanup, clearer problem space definition, and setting stronger boundaries for AI use within the product.
That alone would have saved time by avoiding unrealistic expectations. Pressure often exists to move fast and showcase AI, but without the right foundations, you iterate more than necessary — or worse, lose user trust.
So before jumping in, leaders need to ask: "Do we have the data, the operating model, and the product discipline to take advantage of this?"
How AI is transforming the product-development process
One of the biggest changes we’ve made in the last year is shifting from static, document-heavy product development to AI-assisted, continuous discovery and delivery.
Traditionally, our process relied on structured artifacts — PRDs, business cases, research docs — which were time-consuming to produce and often quickly became outdated. We introduced AI directly into that workflow to synthesize customer feedback, analyze support tickets, generate initial product requirements and even challenge assumptions during discovery.
This changed the speed and quality of decision-making. Product managers are no longer starting from scratch; they’re starting from insight. We’re seeing faster cycles from idea to validation, better alignment across stakeholders, and more data-informed prioritization.
It also changed how our teams operate. Instead of AI as a separate tool, it’s now embedded in daily workflows, whether that’s generating hypotheses, summarizing data, or accelerating delivery. The role of the PM has shifted slightly from creator to curator and decision-maker.
Product managers are no longer starting from scratch; they’re starting from insight.
How product discovery can be redesigned with AI
Out of the whole product lifecycle, though, discovery needs the biggest redesign.
Historically, discovery was slow, manual, and relied heavily on small samples — interviews, surveys, limited data analysis. With AI, that model no longer holds up. We can now process massive amounts of customer feedback, behavioral data, and signals in near real time.
But most teams are still running discovery like it’s 2015.
We started by embedding Jira Product Discovery into our workflow to better structure ideas, insights, and prioritization. That gave us a solid foundation, but we truly accelerated when we layered in Rovo AI agents and connected them to our broader ecosystem — documentation, technical designs, Figma, and other delivery tools.
Additionally, we integrated continuous data streams — survey results, customer feedback loops, contact center data, and product analytics — feeding everything into a single discovery flow.
Now, instead of manually stitching inputs together, teams can quickly synthesize signals, generate initial requirements, and connect discovery directly to execution. The gap between “insight” and “delivery” has narrowed significantly.
This results in faster cycles and better quality. Teams start from a more informed baseline, with stronger alignment across product, design, and engineering.
Why AI informs product activities, but humans decide
We use AI aggressively for scale, speed, and pattern recognition — discovery, data analysis, and early ideation. It’s now our default starting point, not a nice-to-have.
AI synthesizes customer feedback, support tickets, and behavioral data in minutes, shaping hypotheses, and even generating first-pass requirements. In our smart city and digital twin work, we also use it to simulate scenarios, helping teams explore planning and feasibility decisions before building anything.
But we’re very clear on where AI stops.
Prioritization, roadmap decisions, UX nuance, and tradeoffs remain human because that’s where context, experience, and accountability matter. AI doesn’t understand strategy, politics, or risk in a meaningful way. It can inform decisions, but it shouldn’t make them.
For example, we used AI to analyze customer feedback and support tickets to guide prioritization. The model identified patterns and surfaced frequently mentioned issues effectively, but it couldn’t distinguish between what was loud and what was important. Low-impact issues were overrepresented simply because customers reported them more often, while more critical but less frequent problems were underweighted.
It also missed context about the customer journey — why something was happening, not just that it was happening. Human judgment was critical to interpret the signal correctly.
From a trust perspective, if AI surfaced something that didn’t align with stakeholder intuition, stakeholders quickly dismissed it. So accuracy alone wasn’t enough — it had to be explainable and aligned with real-world context.
Why product managers need to be careful of over-reliance on AI
Speed and throughput are the biggest positives of AI. Product managers move from idea to validated direction significantly faster; what used to take days of synthesis and drafting now happens in hours. Across teams, I’d estimate a 30–50% speed increase in early-stage discovery and documentation cycles.
Quality has also improved. AI helps teams start from a more informed baseline, pulling in customer signals, highlighting patterns, and reducing blind spots. This leads to better initial hypotheses and more data-informed discussions with stakeholders.
On the flip side, over-reliance is the biggest risk I’ve seen. Some teams initially treated AI outputs as “answers” instead of inputs, which led to shallow thinking or generic solutions. We had to actively coach teams to challenge and refine what AI produces instead of just accepting it.
A consistency gap also exists. AI is only as good as the context you give it. Strong product managers get exponentially more value because they know how to guide it. Less experienced ones can struggle or produce noisy outputs.
Net-net, the upside is real. Faster cycles, better starting points — but only with strong product thinking. Otherwise, you just get to the wrong answer faster.
Why end-to-end automation is disappointing PMs
Early on, I assumed AI could meaningfully drive roadmap decisions or produce high-quality, ready-to-ship solutions. In reality, what it generates is often directionally useful but still generic. It lacks the business context, customer nuance, and the tradeoffs experienced teams make every day.
End-to-end automation is another area where it hasn't fully landed. AI agents show much promise for running workflows or replacing parts of delivery, but in practice, it still requires heavy human oversight to be reliable at scale.
The gap, then, is between expectation and reality. AI is incredibly powerful as an accelerator, but it’s not a replacement for product judgment — and teams that treat it that way tend to get disappointed pretty quickly.
Why PMs must shift from control to confidence with AI
Product quality no longer comes from tightly controlled, deterministic experiences. We have to let go of the notion that it does.
Traditionally, we designed products to behave predictably — clear flows, defined outputs, edge cases mapped out. With AI, especially generative AI, you’re introducing variability by default. The same input might not always produce the exact same output, and that changes how you think about quality.
It forces a shift from “Is this perfectly defined?” to “Is this reliably useful and within acceptable bounds?”
That’s a big mindset change. You’re no longer designing every outcome — you’re designing systems with guardrails, feedback loops, and ways to continuously improve. It also means embracing a level of imperfection, as long as the value is there and the risks are managed.
So, I shifted from control to confidence—trusting the system to operate within boundaries, rather than trying to define every possible outcome upfront.
How AI alters the dynamics of product teams
AI hasn’t changed the structure of the team as much as it’s changed how the team operates.
We haven't replaced roles, but the expectations of those roles have shifted. Product managers are spending less time on manual synthesis and documentation, and more time on judgment, prioritization, and driving outcomes. The bar for thinking has gone up.
It’s also blurred some of the lines between roles. PMs are getting closer to data, designers are using AI to iterate faster, and engineers are leveraging it to accelerate development. The collaboration loops are tighter because everyone can move faster individually.
At the same time, we’re seeing a clear gap emerge: people who know how to work with AI versus those who don’t. The teams that lean in are significantly more productive.
How building AI into workflows improves adoption
Adoption has been a challenge. Not everyone moves at the same pace. Some lean in, experiment, and adapt quickly. Others hesitate, question it, or wait for it to be perfect before engaging — and that’s where the gap shows up.
We made AI usage visible and expected — not optional.
Instead of rolling it out as “here’s a new tool,” we embedded it directly into team operations. For example, we updated our product workflows so every discovery or PRD cycle included an AI-assisted first pass — whether synthesizing customer feedback, generating initial requirements, or outlining hypotheses. It became part of the process, not an add-on.
We also set a simple expectation: Don’t come to a review or discussion with a blank page. Start with something AI-assisted, then show how you’ve refined it. That shifted the mindset quickly from resistance to practical use.
At the same time, we highlighted wins — showing how one team cut discovery time in half or improved their initial thinking quality using AI. That peer visibility created momentum.
It wasn’t about forcing adoption; it was about normalizing it in the workflow and making the benefits obvious through real examples.
Don’t come to a review or discussion with a blank page. Start with something AI-assisted, then show how you’ve refined it.
Why product leaders must lean into AI — now
Here's my advice: Don’t wait for clarity — lean in and start now.
This is a moment where early movers will define what comes next. You can either sit back and analyze it, or get your hands dirty and figure it out in real workflows. A gap is already forming between those using AI every day and those still talking about it.
Be an evangelist. If you believe this is the future, then act like it. Put your money where your mouth is—use it in your own work, challenge your teams to use it, and build it into how your organization operates. This isn’t about running pilots on the side, it’s about changing the way you build products.
At the same time, stay grounded. AI isn’t a replacement for thinking. It’s a multiplier. Winning leaders will combine strong product judgment with a willingness to learn fast.
This really comes down to learning. If you’re not actively learning, experimenting, and evolving right now, you will fall behind. It’s that simple.
So lean in, stay curious, and lead from the front.
Follow along
You can learn more about Tony Pagliocco's work at tonypag.com or connect with him on LinkedIn.
More expert interviews to come on The CPO Club!
