Nodo
Year
2026
TYPE
Mobile
CONTEXT
Artificial Intelligence
DURATION
6 months
OVERVIEW
A conversational AI layer for ecommerce, designed as a personalized shopping assistant that enables users to discover, compare and select products through natural language, integrated into mobile apps as a scalable, multi-brand solution aimed at increasing conversion while introducing a more adaptive and context-driven shopping experience.
CHALLENGE
Conversational interfaces change the interaction model of ecommerce at a fundamental level – replacing structured navigation with open-ended, intent-driven input. That shift unlocks flexibility and personalization, but it comes at a cost: more ambiguity, less predictability, and a heavy dependence on data quality. A user can ask for almost anything, in almost any phrasing, and the system still has to respond usefully.
Nodo compounded that challenge by living inside a SaaS ecosystem, where a single solution had to work across multiple clients, product categories, and brand identities at once. The core design problem was twofold: interpret genuinely diverse, unpredictable user inputs and translate them into relevant product suggestions – while holding consistency, usability, and visual adaptability across every brand it plugged into.
PROCESS
Ran usability testing with ~30 participants across interactive prototypes and live environments, measuring task completion, input behavior, and how users' expectations shift when they're interacting with an AI rather than a conventional interface.
Structured the system around open input, pairing AI-generated responses with product recommendations and follow-up prompts that progressively refine what the user actually wants.
Designed the states most conversational products neglect: empty states, error handling, and fallback scenarios that keep the experience continuous even when the AI hits its limits.
Built the interface for visual neutrality, so a single system could integrate cleanly into the distinct design languages of different client brands.
Worked with product and engineering on scope, feasibility, and implementation – including the architectural shift from webview to native mobile.
SOLUTION
Nodo adds a conversational layer inside ecommerce apps, letting users search, explore, and select products through natural language. It opens from a floating entry point into a dedicated interface where the user engages directly with the assistant ≥– always one tap away, never in the way of the existing shopping flow.
What makes it work is that the conversation stays actionable: text responses come paired with dynamic product cards, so users can view details, add to cart, or save an item without ever leaving the chat. And every exchange does two jobs at once – it answers the immediate request and moves the user forward, using follow-up prompts to refine intent and support a genuinely iterative path to a decision, rather than a single query that either hits or misses.
IMPACT
Nodo launched at VTEX Day 2026 as part of the company's innovation initiatives, where it drew strong client interest and served as a flagship demonstration of applied AI within the platform. As an MVP, it marked the first conversational experience introduced into the SaaS ecosystem – and the appetite for it was immediate: during launch, more than 50 clients expressed interest, including major Brazilian brands across sectors like health and fashion.
That interest converted into real adoption. Within the first months, Nodo was running live across 21 client apps and had generated over 30,000 sessions, with 43 additional clients queued for rollout in Q3 2026. Beyond the numbers, the project set the foundation for how the company approaches conversational commerce – shaping product direction and technical exploration going forward, while deepening my own experience designing systems that hold up across multiple brands, categories, and levels of content maturity.
