The shift to autonomous resolution

For years, the customer service chatbot has been an exercise in frustration. Most legacy bots were interactive FAQs designed to prevent customers from reaching a human agent by offering a menu of predefined options. This defensive posture saved costs but often degraded the user experience. Eli, a Paris-based startup founded in 2023, is part of a new cohort of companies attempting to flip this script. Instead of merely deflecting tickets, Eli builds autonomous agents designed to resolve them.

The difference between deflection and resolution is the core of the current AI agent narrative. While a standard chatbot might tell a customer where to find the return policy, an agent like Eli is granted the authority to actually process the return. This requires more than just a large language model; it requires deep, two-way integration into the e-commerce stack. Eli focuses primarily on the Shopify ecosystem, connecting to help desks like Gorgias and Zendesk to act as an automated member of the support team rather than just a front-end filter.

Specialized architecture for retail

Eli's technical approach relies on agentic workflows—a series of loops where the model can check information, verify user identity, and then execute actions within a brand's existing tools. For a high-growth apparel retailer, this means Eli can handle the bulk of "Where is my order?" queries that typically bog down human staff. By connecting to Klaviyo for customer data and Shopify for inventory and order management, the agent provides personalized updates and initiates changes that generic bots cannot.

The company sits in a competitive but lucrative segment. Its most direct competitor is Sierra, the agent platform co-founded by Bret Taylor. While Sierra targets massive enterprises with complex, custom needs, Eli has found a niche in the mid-market and high-growth retail space. It also competes with Intercom’s Fin, though Eli’s advocates argue that being a specialized, agent-first platform allows for tighter integration into retail-specific logic than a horizontal platform like Intercom.

Outcome-based labor

Founded by Guillaume de Grave and Benoît Terpereau, Eli is a product of the growing AI cluster in Paris. Unlike high-profile foundational model companies like Mistral, Eli is focused on the application layer. Their bet is that the value of AI lies in specialized execution rather than general reasoning.

The pricing model reflects this focus on outcomes. Rather than charging per seat or a flat monthly fee, Eli often operates on a per-resolution basis. This aligns the company's incentives with the brand's: the agent is only profitable for the client if it actually solves the problem without human intervention. This shift toward outcome-based pricing is a hallmark of the burgeoning agent economy, signaling a move away from the SaaS tax toward a labor-as-a-service model.

Despite the promise of autonomy, Eli faces the hurdle of API reliability. An agent is only as good as the tools it can access. If a brand's Shopify instance is poorly maintained, the agent's effectiveness drops. Eli’s long-term success depends on how well it can normalize this data and provide a plug-and-play experience for brands lacking large engineering teams. As the agent stack matures, Eli’s position as a specialized, action-oriented layer for e-commerce makes it a notable player in the transition to autonomous service.