AI for sellers: what Olist's move reveals about the future of multichannel operations
Olist put AI at the center of its strategy and grew net revenue by 65%. For multichannel sellers, the signals are clear: autonomous agents, automated customer service, and proprietary models will separate those who scale with real profit from those stuck in disconnected data.
AI agents that close sales on their own, 100% automated customer service, and a platform growing 65% a year — Olist is becoming an AI-native company. And this isn't a distant curiosity for those selling on marketplaces. It's a sign of what's coming for multichannel sellers who still operate manually.
In practice, Olist already uses artificial intelligence to qualify leads, close lower-complexity contracts, and manage first contact with customers in the logistics division. About 20% of sales are closed end-to-end by autonomous agents. Primary customer service is almost entirely handled by Lis, an orchestrating agent running on multiple LLMs. For a seller operating on marketplaces like Mercado Livre, Shopee, and Amazon, the message is direct: intelligent automation is shifting from a differentiator to a survival requirement.
What changes in practice for sellers
Olist was born as a management platform (ERP) for small and medium businesses. Today, it's an ecosystem that includes marketplace integration, financial services, and logistics. The pivot to AI isn't cosmetic: it changes the speed and precision of decision-making. For a multichannel seller, this means tasks like answering customer questions on WhatsApp, analyzing the real margin of an order, or deciding how much to invest in ADS can be delegated to agents that learn from operational data.
At Jodda, we see this move as the consolidation of a category we call profit intelligence. It's not enough to have a dashboard with numbers. The platform must pinpoint, order by order, where the real profit lies — considering fees, shipping, commissions, and taxes from each marketplace. Olist is investing in proprietary models to lower the cost of these analyses and make them deeper. This is an indicator that the market is moving toward solutions that deliver decisions, not just data.
AI agents and order-level profit
Olist's CEO, Tiago Dalvi, told Bloomberg Línea that the cost to execute an action via AI has dropped about fivefold in the last 18 months. This makes mass use of agents viable for tasks that previously required a human — like checking if a marketplace payout is correct or calculating the real margin of a SKU with subsidized shipping.
For multichannel sellers, payout auditing is one of the most sensitive points. Different marketplaces have distinct commission rules, fees, and subsidies. An AI agent can cross-reference this information with product costs and deliver a real profit reading per order, something manual spreadsheets rarely achieve without error. Jodda already operates on this logic: we centralize orders, costs, and taxes to show the real margin, and we see AI as an ally to scale this intelligence without losing precision.
What Olist's data shows
According to the report, Olist has just under 60,000 active customers and more than half already use AI features in their journey. About 30% use these tools on a recurring basis. These numbers indicate rapid adoption, but also a growing gap: sellers who ignore AI will fall behind in operational efficiency.
Another relevant data point: Olist's average ticket rose nearly 50% year-over-year. This suggests that automation isn't just cutting costs — it's allowing the company to move up the value chain and serve larger customers. For a seller, the lesson is clear: using AI for repetitive tasks frees up time to focus on strategy, supplier negotiation, and channel expansion.
Proprietary models and the future of profit intelligence
Olist plans to create proprietary AI models in the coming years, aiming for cost reduction and defense of its business intelligence. This is a trend that should spread: platforms serving multichannel sellers will develop AI layers trained on specific Brazilian retail data — taxes, marketplace rules, seasonality. This could make the generic profit analyses that many sellers still do in Excel obsolete.
At Jodda, we believe profit intelligence is the layer that transforms raw data into action. It's not about replacing the seller, but giving them a clear view of the real margin, order by order, so they can decide where to invest, which channel to prioritize, and how to adjust prices. Olist's move reinforces that this is the market's direction.
What to do now
For multichannel sellers, now is the time for experimentation. Start by testing simple automations: answers to frequently asked questions in post-sales, alerts for orders with negative margins, ABC curve reports by marketplace. The technology is cheaper and more accessible than ever. The risk isn't in adopting AI too early, but in arriving late to a game already being played by more agile competitors.
Olist is becoming an AI-native company. The multichannel seller who wants to compete in 2026 and beyond needs to start building their own profit intelligence — with or without a platform, but never without criteria.