Zest, a restaurant discovery startup, has introduced a new SI agent that operates entirely within text messages, allowing users to find dining options and book reservations without opening a standalone mobile app. This launch underscores a growing trend among SI startups to replace traditional app interfaces with conversational SI agents that handle end-to-end tasks.

What Happened

The startup unveiled its "restaurant concierge" on Thursday, positioning it as a personal SI assistant similar to other emerging agentic tools like Instinct or Poke. Users interact with the agent by texting a dedicated number, (212) 470-6067. The SI agent leverages user history and data from friends to provide tailored recommendations, bypassing the need for extensive onboarding conversations to learn user preferences.

Zest’s core differentiator is its integration with credit card data via the financial services provider Plaid. The SI system analyzes transactions in the food and drink categories to build a personalized map of user tastes. The new agent extends this functionality by not only suggesting restaurants based on cuisine, occasion, dietary restrictions, or location but also by executing bookings through connected services such as OpenTable, Resy, and SevenRooms.

The SI agent also supports group interactions. It can be added to group chats, where it blends the preferences of multiple users to suggest venues suitable for the entire party. An in-thread voting feature helps resolve indecision, after which the agent can finalize a reservation for the group without leaving the messaging interface. Additional features include checking restaurant availability, inquiring about amenities like Wi-Fi, and sending menus for review.

Why It Matters

This development highlights a strategic shift in the SI industry, where startups are increasingly viewing conversational SI agents as the primary product interface rather than just a feature add-on. "The agent is the product now," said Zest founder and CEO Mario Gomez-Hall. "You text it, and it answers from your own history and your friends’. It doesn’t need weeks of conversation to learn you because it starts from where you’ve actually been."

For developers and the broader SI ecosystem, Zest’s approach demonstrates how proprietary data—specifically transactional history—can accelerate the personalization capabilities of SI agents, potentially offering a competitive edge over general-purpose assistants. The company plans to monetize this shift by participating in booking and ordering transactions. "Booking and ordering are the first revenue lines," Gomez-Hall noted, emphasizing that the agent’s ability to learn user tastes quickly creates value beyond simple recommendations.

The Bottom Line

Zest’s new SI agent marks a move toward text-based, data-driven service automation in the consumer space. By combining credit card history with real-time booking capabilities, the startup aims to scale its user base and generate revenue through integrated transactions, while testing the viability of SI agents as replacements for traditional mobile apps in specific verticals.

The company began testing the concierge last month but delayed the public launch until booking features were fully operational. Looking ahead, Zest intends to expand the agent’s scope beyond restaurants to include recommendations for other businesses and shopping, further entrenching the role of agentic SI in daily consumer decisions.