Ailo: from experiment to product with purpose

Over the past few months, Ailo, our AI agent for consumers, has evolved from a small experiment into a product with purpose.

If you missed our previous posts where we shared the creation process of this AI agent, we recommend checking out this and this article first — there we talk a bit about the process and what happens behind the scenes.

Clayton Christensen, in The Innovator’s Dilemma, shows how established companies often get stuck in incremental improvements, while major transformations are born from experiments that seem small at first.

Ailo started this way, almost invisible, and now shows signs that it can change how people order on iFood.

This article is about this turning point: the problems we’ve solved so far, the first signs of product-market fit, and, most importantly, how people are receiving this product.

First signs of product market fit

Today, Ailo is available in the iFood app in 5 cities, and on WhatsApp, for anyone in Brazil.

The flow seems simple: users chat with the agent, which presents a carousel of items based on order history and conversation context. Then, the journey continues in the iFood app, in the experience that already exists today.

But behind this simplicity, the numbers stand out: Ailo users create 2x more carts than in the traditional app experience. Additionally, our research shows that 80% of users find Ailo helpful in moments of indecision.

“iFood has so many options, like Netflix, and the assistant helped me find a shorter path.” — user comment

And this is where the heart of our product lies: ensuring that this task, which seems simple but is extremely complex behind the scenes, is done in the best possible way for users.

Data and people: the heart of the product

The hardest part of building a recommender isn’t the model itself, but the quality of the data foundation. None of what we’ve shown here would be possible without iFood’s structured data foundation and the LCM (Listing Candidate Manager), which ensures that user intent is always aligned with available offerings in real time.

But data alone isn’t enough. Ailo is a two-way street: it learns from each interaction, improves based on user feedback, and continuously refines itself thanks to the combined work of people, data, and technology. It’s in this combination that the experience truly comes to life.

“I wasn’t interested in any of the suggested restaurants. There wasn’t any restaurant I knew.”

user comment

We’re still learning to balance two important moments: exploring new options and the comfort of surfacing what users already know and love. This has been our current challenge and every piece of feedback is essential!

Latency is every agent’s Achilles heel

If you work with agents, you’ve probably faced the latency problem. For those unfamiliar: latency is the time it takes for the system to respond after you ask a question. It’s like when someone takes too long to respond in a conversation — the flow is lost and impatience takes over.

In the beginning, our latency reached 16 seconds or more to generate results, which significantly compromises the experience, especially when users are hungry and possibly irritated. Through research and constant monitoring, we were able to identify this pain point and prioritize improvements until we reduced the time to acceptable levels.

We made two important moves:

• We adjusted how Ailo processes each message. Simple requests, like a “hi”, don’t trigger deep searches, ensuring faster responses.
• We made the architecture more dynamic, showing micro-interactions during complex searches so users can see that Ailo is working in the background.

This balance brought more naturalness to the conversation and more confidence in usage.

The channel shapes behavior

Another important learning was realizing that the channel directly influences behavior.

• In the app, less than half of users perform more than one search. This happens because the comparison is always with iFood’s traditional experience, built over the years.
• On WhatsApp, the dialogue is more natural and continuous, closer to a real conversation.

These differences show that there’s no single “Ailo user” — there are distinct journeys, shaped by the channel and context. We’re now exploring new ways to highlight Ailo within the app experience, respecting expected behavior and iFood’s current journey.

The magic is in the practicality

We’re increasingly noticing that Ailo is becoming a relief for users by recommending according to their profile and speeding up the purchasing process.
People don’t want “AI”. They want its benefits — saving time, saving money, solving a problem.

This echoes Theodore Levitt’s iconic phrase: “People don’t buy drills, they buy holes in the wall.”

“If it were from scratch, it might be different. But since it was adapted to my profile, it was fantastic.” — user comment.

The real magic lies in practicality and the feeling that someone understands your needs at the right moment.

What’s next?

We’re just getting started. But we already have clarity on some directions:

First, gradually expand and make Ailo available in the iFood app. We’re eager to see people from different regions chatting with Ailo inside the app. Meanwhile, we continue gathering feedback about the WhatsApp experience and iterating.

Another discussion is making personalization even deeper: we’re exploring new moments where Ailo can fit into people’s daily lives, becoming increasingly proactive, always with care, responsibility, and security.

Agent products are still finding their place in the market, and this future is constantly changing. But our vision is clear:

“Ailo will understand our needs at the right time, with learning so fast it seems like magic, but it’s the power of AI!”

Bruna Faim, Ailo product designer

“Ailo’s future is to be an assistant that understands you as well as a friend and anticipates your desires.”

Renato R. Ribeiro, Ailo Software Engineer

“Ailo will manifest in a multimodal way, understanding context to choose when and where to be more visual, conversational, or interactive.”

Camila Gargantini, Ailo product designer specialist

“Ailo will be like a friend who knows what I want and places the order for me.”

Chiara Caratelli, Ailo Data Scientist

“Ailo isn’t just a new way to order food, but a virtual iFood assistant that understands me like no one else.”

João Lippi, Ailo Software Engineer

“Ailo’s future is to make my routine easier: planning my groceries and meals, with voice commands, everything my way, effortlessly.”

Lucas Mattos, Ailo GPM

If at first it seemed like just another experiment, today we already see Ailo as a fundamental part of iFood’s future. And this future is being built with every conversation, every choice, and every order placed with our agent’s help.

👉 We invite you to chat with Ailo on WhatsApp, click here and try it now!

And following our culture of collaboration and testing, send us your feedback here if you find any issues, improvement suggestions, or something you liked about the experience.

See you next time! 👋

Article written in partnership with Valéria Romano, Senior Product & Design Manager.